I sat down with Michael Anderson, Chief Product Officer at Filevine, at the LEX Summit in San Diego, where attendance topped 2,000 this year. I wanted to start with his “Back to Brains” article, which argues that lawyers spend too much time pointing, clicking, and typing to get software to do what they already know they want. Filevine announced the LOIS pen and notebook at the conference. It is a real pen with real ink on real paper, and a sensor sends the handwritten notes into the meeting transcript, so a litigator in a client meeting or a focus group can keep eye contact and still capture the thought. Michael calls the legal professional the hero of the story. That is also why LOIS has dictation and readout, since he says the best case-strategy ideas tend to show up when you’re on the go.

The law librarian in me had to ask about the citator. John Rizner demoed it for us earlier (Marlene and I assumed it would go over most people’s heads, until two partners from my firm told me it was the greatest thing they had ever seen). Case results get a green check mark when the citation is verified, or a magnifying glass when something needs a second look. Claims about what an opinion says link to the passage that supports them. You can then highlight a specific holding and run what Michael called a semantic Shepardization of that text across the full corpus of U.S. case law. It uses machine learning to look for similar language, so it can surface cases that address the same point without citing the original, including sub silentio treatment. For people outside the profession, Michael uses Spotify to explain it. Most of us let it recommend songs based on what other listeners picked, and sometimes you want to ask for pop songs with a cello.

We also talked about acquisitions. Filevine bought a deposition marketplace that schedules digital court reporters and paired it with its own live deposition feedback tool, which flags when a deponent says something inconsistent with earlier testimony or with a document in the case file. Pincites brings the work into Microsoft Word. Michael’s reason litigators pick these tools up is that “intelligence is always competitive,” and one fact your opposing counsel doesn’t have gives you a chance to win. He also said something that matches what I see at my firm. Eight years ago he ran SaaS implementations where the main job was getting staff to use the system, and he has been surprised at how quickly attorneys have taken to AI by comparison. He credited Jack Newton with the line that LLMs are to attorneys what spreadsheets are to accountants.

The headline announcement was voice agents that answer inbound calls to the firm. In the live demo on stage, the agent took a call, identified the caller, followed the firm’s decision tree, and put the data into the intake system. Setup uses what most firms already have, such as the script you would hand a new front-desk hire or the configuration in a system like Lead Docket. When I asked about guardrails (I still think about the Air Canada chatbot), Michael said the agent works from the same process documents a new intake employee gets, and that employee doesn’t have full case access either. He told a story about calling several law firms about a real estate matter and getting voicemail every time, with no callbacks. He also pointed to the California State Bar’s Justice Gap Study, which shows how many Californians with legal problems never get the help they need, and he sees the after-hours, overflow, and voicemail calls as the place to start.

We covered how Filevine handles the pace of releases. LOIS now supports external connectors, including Outlook and document management systems, and partners want to bring their own MCP servers into the platform. Michael said an orchestrator picks the right tool for each request, so users get new capabilities without hunting for a menu item. When I asked him to look ahead, he went back to brains. AI does well at drafting a brief when the person using it knows how to draft a brief. He sees his own middle schooler and high schooler being taught skills the analog way before they get the new tools, and he thinks legal will need the same approach: learn the skill first, then use the technology to speed it up.

Listen on mobile platforms:  ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Apple Podcasts⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ |  ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Spotify⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ | ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠YouTube⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ | ⁠Substack

Email: geekinreviewpodcast@gmail.com

Music: Jerry David DeCicca

Transcript

Greg Lambert (00:07)
Welcome to The Geek in Review. I’m Greg Lambert, and this is the podcast on creativity and innovation in the legal space. Today I am in beautiful San Diego, California, at the LEX Summit for Filevine, and I have with me Michael Anderson, who is the Chief Product Officer at Filevine. We had his brother on earlier this year, along with John Rizner, but

Greg Lambert (00:36)
welcome, Michael. It’s good to have you.

Michael Anderson (00:37)
Great to be here, Greg. Thank you.

Greg Lambert (00:39)
So I know on the conference, I think you’ve topped two thousand attendees, roughly, and that’s quite a rise from last year. Pulling it all together, is everything going the way you were hoping?

Michael Anderson (00:56)
I think so, yeah. We have such an engaged community among our customers and prospective customers and partners. It’s really great as a product guy to be able to put something in front of them and know that they’re going to try it out, right? That’s just like a gift. It’s really great to have that. So I love their engagement, love their energy. It’s been really good so far.

Greg Lambert (01:18)
Yeah, it looks like an enthusiastic crowd. I want to talk to you about something that you wrote. It’s a “Back to Brains” article, and in it you were talking about how you were working to get away from the pointing and clicking and typing and all of the things that you have to do just to get the tool to do something.

Greg Lambert (01:46)
Do you mind expanding on your thought process there?

Michael Anderson (01:49)
Yeah, a couple of things I can say about that. So one, as it relates to today, we announced the LOIS pen and notebook. This is a real pen with real ink writing on real paper, and it can do a couple of things. We have a whole LOIS for Meetings tool where you can have a meeting like this recorded, and we know that, especially for

Michael Anderson (02:17)
litigation-focused groups, there are lots of highly consequential conversations. And having a screen intermediating those conversations is not going to do it. Sometimes you really need to be in the room. You need to maintain eye contact to move someone in a particular direction. But along the way, you may have brain synapses fire and remember something that you want to follow up on, or you may notice something that

Michael Anderson (02:46)
the person on the other side of the table did, and you make quick notes of it, right? Pen-and-paper type of notes. So what we did is we partnered with a smart pen company. Again, it’s real paper, real pen, but it has a sensor on it, and all of those notes end up going into the transcript of the meeting. So your human soft skills that you have as a litigator, as an interrogator,

Michael Anderson (03:14)
or as someone who’s developing a relationship with a client or an expert witness or whatever it might be, we say those things are really, really good. We need to do more to support those things and have technology adjust around them rather than the other way around. So I would say that’s relevant to today because we announced this LOIS pen. We gave it to everybody here attending LEX, and we’re excited about that.

Michael Anderson (03:44)
But then I would say part of our philosophy is that the legal professional is the hero of our story. Whether it’s giving them a pen and paper to use in a highly consequential meeting or allowing them to be in a high-agency mode, they already know what they want to have happen. Just make the technology responsive to what they already know,

Michael Anderson (04:14)
and use those neural pathways that they’ve developed over years or decades to drive the technology, and not the other way around.

Greg Lambert (04:24)
Yeah, it’s really interesting with the pen and paper, because I think a lot of us are still getting used to the fact that you can talk to the software rather than type. And I just wonder how many lawyers will have to relearn how to write with pen and paper, because we’ve been typing for the past twenty-five years.

Michael Anderson (04:48)
I’ll say this,

Michael Anderson (04:49)
Greg, as well. Eight or nine years ago, I spent a lot of time inside law firm cubicles, teaching people to adopt a new system of record, getting them to go through the click navigation path. And I actually just recently had to engage with a new system for something, and talk about cortisol rising.

Michael Anderson (05:16)
You get in there, you see something for the first time, and you’re like, I don’t even know where to click. I’m told it has all these capabilities, but how do I get to them? So the fact that you can engage with technology through spoken voice, through pen and paper, right, that’s part of our vision. That’s why LOIS has dictation and readout. Some of your best ideas in terms of case strategy, or when you’re working on a particular deal,

Michael Anderson (05:45)
are not going to come to you while your fingers are on a keyboard, staring at a screen, right? They’re going to come to you while you’re on the go. You need to be able to dictate, memorialize them, and bounce them against a sounding board. That’s exactly what people can do in LOIS, and that’s exactly what people expect in the AI era. And so I think that change in interface puts the onus for the thought work back on the practitioner.

Michael Anderson (06:15)
So anyway, yeah.

Greg Lambert (06:16)
As much as possible. The more you can keep hands off the keyboard and off the mouse.

Michael Anderson (06:21)
Yeah, exactly.

Greg Lambert (06:24)
I had talked earlier with John Rizner, who was on the podcast earlier with Ryan. Being a law librarian by training, and having learned research skills over the past thirty years, I was really excited when John was demoing the citator that you have. And I know this is something that

Greg Lambert (06:54)
your audience or your customers may not know exactly. Do you mind walking through what the citator is, what it does, and where you see the value?

Michael Anderson (07:05)
Yeah, absolutely. For our customers, especially if you’re looking for court opinions going from the state supreme court level all the way to the top, we have a full corpus that is really best in the industry. It’s the same corpus that’s used by Google Scholar. And what we allow you to do is certainly just chat and find

Michael Anderson (07:34)
relevant case law. But after you do that, there are a couple of things you’re going to want to do. You’re going to want to know how much of what just came back is slightly off, or maybe entirely hallucinated. So on the first answers you get back from our corpus of legal opinions, we provide visual indicators. If it’s fully verified, meaning the court opinion

Michael Anderson (08:04)
is as it is written and is cited as it should be, there will be a little green check mark next to it. If it might need a bit more investigation, there will be a small magnifying glass that says, hey, check this, maybe it’s mislabeled, maybe it’s totally off. So we provide that visual indicator for the case names themselves. Then we take it a step further in two different ways. You get full pincite-level

Michael Anderson (08:34)
granularity, so the claims the AI comes back with about what a legal opinion says are grounded in the words and phrases of the actual opinion. That’s one. And then once you decide, okay, I want this holding, and I need to know if this specific text is still good law, we run a full semantic Shepardization

Michael Anderson (09:02)
of that highlighted text against the entire corpus. And let’s be clear, this is not Shepardization that is done by hand, and that’s on purpose. This is Shepardization targeted to specific paragraphs, the specific things you highlighted, and it’s looking for semantic similarities. There’s a lot of machine learning work that goes into this.

Michael Anderson (09:30)
It’s looking for semantic similarities across the corpus of law, meaning you’re going to find things that are directly cited by others, but you’re also going to find sub silentio treatment that isn’t directly cited, because it’s not done by hand. It’s done with deep semantic retrieval across that corpus. I don’t know if there’s any part of that you want to dive into, but that’s how I would summarize it.

Greg Lambert (09:54)
I think one of the things people may not understand is that, traditionally, a Shepard’s or a KeyCite looks at cases that have cited other cases on a particular issue, but there really need to be two things going on at the same time. There needs to be a direct cite to the case, and then the topic for which it’s citing that case is captured, and they correspond. And if you’re missing one of those two, it doesn’t

Michael Anderson (10:21)
The web’s incomplete.

Michael Anderson (10:22)
Yeah.

Greg Lambert (10:23)
show up. And so by using this kind of knowledge graph that you’re doing with the citator, if it’s the same topic, even though it may not cite that case, if they’re talking about the same thing, it overlaps, and then you can verify that. Is that right?

Michael Anderson (10:39)
Yeah.

Michael Anderson (10:39)
Yeah, I think that’s exactly right. Sometimes I explain this to internal folks in marketing or engineering, and I’ll ask, “Do you use Spotify?” And they say yeah. And I say, okay, Spotify will give you songs it thinks you’re going to like based on what other people have chosen. But what if sometimes you say, actually, I don’t want you to tell me what you think I’ll like or what might be relevant. I want to match a particular sound.

Michael Anderson (11:08)
I want pop songs with a cello, or whatever. I use that as a metaphor for people who don’t have a lot of exposure to this. Yeah, we’re looking for text that is similar across the corpus of United States case law.

Greg Lambert (11:24)
Yeah, it was funny, because when John was describing this, myself and another librarian, Marlene, my co-host, were going, “This is great, but it’s going to be way over the heads of everybody here.” And then I talked to two partners at my firm who were there, and they said, “That was the greatest thing I’ve ever seen in my life.” So it does seem to strike a chord. I want to move on.

Greg Lambert (11:49)
You’ve done some acquiring over the years, and just like you were talking about, you want to get lawyers away from the typing, clicking, pointing, and moving things around. But you also want to be there in the products they’re

Greg Lambert (12:08)
actually using at the time, so they don’t have to context switch. You’ve had a couple of acquisitions. You had Pincites, which you’re using in the Microsoft Word environment, and then Parrot, which works in the deposition space, the copilot for depos. Talk about why you made those acquisitions, how they’ve integrated into Filevine, and

Greg Lambert (12:37)
how they’ve advanced since they came over.

Michael Anderson (12:41)
Yeah, great question. We’ve made some great acquisitions over the past couple of years, so I’ll start in chronological order. On the depositions front, we acquired a company that described itself as the Uber of depositions. They had a marketplace of digital court reporters, and you could schedule a deposition through them. They make sure it gets assigned to

Michael Anderson (13:09)
the right court reporter for your jurisdiction. Then you have an entire platform the deposition is run from, with tools for after the deposition, and there was a lot of AI in there as well for taking snippets of the deposition you might want to use in court. We paired that with a live deposition feedback tool that we had already

Michael Anderson (13:39)
released. So you now have an end-to-end experience: pre-deposition, creating a deposition outline, being in the deposition and getting live feedback that, you know what, this deponent is saying something inconsistent with what they said two hours ago, or inconsistent with one of the documents in your case file.

Michael Anderson (14:03)
And then you take the intelligence that comes during the deposition, which you could act on after a break, or after seeing it, or because someone in second chair tells you, and that goes all the way through to getting a delivered transcript. So you wanted to know why we did this. We were deep with litigation customers. We have

Michael Anderson (14:32)
litigation founders. We’ve been pretty darn committed to the litigation space. And depositions are a big part of what it means to be a litigator, to be a litigation firm. You’ve got to have a strong depositions muscle. There’s an industry that supports this, but it’s not doing it in an AI-native way.

Michael Anderson (15:01)
So we made the acquisition, paired it with some technology we already had, and now you’ve got a really cool marketplace and a product experience with intelligent functionality inside it that helps you during the deposition and after the deposition.

Greg Lambert (15:19)
Well, we all know that lawyers, especially litigators, love change, right? I’ve seen this myself. When you introduce this type of change in their process, especially something as core as depositions, you expect a little pushback: why do I need this? How do you

Greg Lambert (15:47)
talk to your customers to explain what the product does and why it makes things better, and then what does the training or implementation phase look like to help push this along?

Michael Anderson (16:00)
Yeah, great question. I think there are a couple of things that really pop out. One is just how easy it is to schedule a deposition. We have customers on both the plaintiff and the defense side, and they may start out booking a handful of depositions with us a week, and then grow to booking dozens or even hundreds. So there’s something to

Michael Anderson (16:27)
the ease of getting something scheduled. But I would say the difference maker is that you’re already in possession of things that could change the game during this deposition, both in your preparation and in your execution. Those are in your evidentiary materials. They’re things you need to retrieve and understand better, and that will help you.

Michael Anderson (16:56)
We help you prepare the deposition outline using the material you already have, and then we reference it during the deposition. I think there’s something interesting about the AI era, which is that intelligence is always competitive. Intelligence is always competitive, right? If you have one piece of intelligence more than your competitors, more than

Michael Anderson (17:25)
your opposing counsel, you have a chance to win. And I think that makes it different. Litigators want to win. So the nature of having a constantly competitive intelligence dynamic in the industry, and not wanting the other side to have an advantage, I think that helps us.

Greg Lambert (17:48)
I like what you said about intelligence being competitive. I get asked a lot how the legal industry is adapting to and accepting artificial intelligence as part of the work. And my response is always that

Greg Lambert (18:13)
typically, when it comes to technology, the legal industry is about five years behind our clients, except for AI. I’m seeing a lot of lawyers and law firms, even huge law firms, jumping in. And my answer to why was that this is

Greg Lambert (18:38)
one of the first times the technology is language-based. It’s dealing with the same thing we deal with. And your point that intelligence is competitive fits along with that. So were you surprised at how quickly the industry jumped on artificial intelligence and started embedding it?

Michael Anderson (19:05)
I

Michael Anderson (19:06)
think your insight that this is language-based is the key. Was I surprised? Compared to how folks adopted, say, a traditional SaaS platform, which, as I mentioned, I was on the front lines of implementing eight years ago and ran a team implementing, typically the job was to get the staff to use it.

Michael Anderson (19:35)
And the attorneys, as long as they got their reports and could access their documents, they were fine.

Greg Lambert (19:41)
Print it out and put it on

Michael Anderson (19:43)
Yeah, exactly. In the AI era, it might be inverted. The ease with which attorneys engage with this has been interesting and surprising. But it does make sense, right?

Michael Anderson (20:03)
It’s a technology that trades in language, and it’s working with a profession that trades in language. One of our fellow leaders in the industry, Jack Newton, said, and credit to him for coming up with this, that what spreadsheets are to accountants, LLMs are to attorneys. And I think that’s a good way to think about it.

Michael Anderson (20:33)
I came from technology in finance. You put a spreadsheet in front of a finance or accounting person, and they know exactly what to do with it. I feel like it’s the same way here. You put it in front of someone who has a facility with language, and they just know what they would ask and what they want to get. So I think your point is spot on.

Greg Lambert (20:52)
Yeah, and you hit on something I hadn’t really thought about, that this technology may be better focused on the lawyer side than on the business professional side. Because I think a lot of the business professional side is

Greg Lambert (21:14)
moving information from one place to another and making sure things are done on time, whereas this goes right to the core of what lawyers do day in and day out. That’s interesting, because I am seeing a bit of a disparity between the staff side and the lawyer side on adoption.

Michael Anderson (21:34)
So

Michael Anderson (21:34)
I think that’s spot on. It can be both, and we actually see that it’s both among our customers. There are a ton of process gains to be had from using LOIS, and we love those. But what feels new and interesting is the substantive gains lawyers experience when using LOIS.

Greg Lambert (21:54)
All right. So we talked a little bit about how you’re using pen and paper and trying to work the way lawyers work. One thing you announced today is that Filevine’s LOIS can now handle inbound calls. You have a voice that can handle those calls and then

Greg Lambert (22:21)
improve the intake process. Walk me through the new voice agents.

Michael Anderson (22:28)
These

Michael Anderson (22:28)
are our... yeah, great question, Greg. These are our voice agents, and we demoed them live on stage. That was all real, for those who were at LEX. Someone placed an inbound call to a law firm, and a voice agent answered. I think we’ve all had the experience of calling a bank or a hospital or whatever, and you often get a voice agent on the other side. What

Michael Anderson (22:56)
I hope stood out to people who heard it was just how competent that voice agent was. Compare that to what a lot of our law firm customers have now in terms of third-party answering services. Those are good too, but these voice agents actually made progress with the person who was calling the law firm. They collected data from them,

Michael Anderson (23:25)
they put that data in the system, and they followed a whole decision tree. The agent identified an inbound caller who fit a particular persona. So there are ways to build this with all of the context our customers have in their intake systems and in the

Michael Anderson (23:53)
operating procedures they run at their front desk. You take that, you give it to a voice agent, and you have a fully connected experience with your front desk and intake systems of record. And you’re actually making progress with inbound callers. Being on the other side of the line calling, you get a sense of that competence. I’ve had the experience

Michael Anderson (24:23)
for my own personal needs. In recent years I had a real estate matter I needed to work out, and I called several local law firms. Voicemail across the board.

Greg Lambert (24:37)
Did you get any calls back?

Michael Anderson (24:38)
No, no calls back. People are busy, right? People are busy. And I would have really liked to know that I was in someone’s system.

Michael Anderson (24:52)
That someone had a good sense of me, that someone had walked me through getting into that system, that an AI had asked me follow-up questions. So it is a mental shift. I’ve talked to people today who say, well, we have this whole intake team that’s dialed in, and we’ve got it working the way we want it to work. And my message to them is, you don’t have to change any of that at all. But think about all of the calls that hit voicemail. How many of those

Michael Anderson (25:21)
are actually in your firm’s system? How many are actually getting followed up on? Think about that. And then think about the calls that go to call overflow, or an after-hours or weekend service. That’s what you need to be thinking about. The technology here has gotten really good, and we wanted to offer it to our customers and give them a way to

Michael Anderson (25:49)
amplify, accelerate, and scale their front desk and intake functions.

Greg Lambert (25:55)
As you were talking, a couple of things came to mind. You have a diverse customer base. A customer could be a personal injury firm, an immigration firm, a variety of things. How much time and training does it take to get the voice agent to understand my law firm versus another person’s law firm? What goes into it?

Michael Anderson (26:22)
That’s a great question.

Michael Anderson (26:23)
The fortunate thing is that a lot of firms have already done this thought work. Like I said, there’s sometimes a script with a decision tree that you would give to a new hire. You may have already put this information in a system like Lead Docket. So the message is, you already do this.

Michael Anderson (26:49)
Take those documents, take that system configuration in your intake or front desk system, and our voice agents can plug directly into it and follow it. There may be some other bells and whistles you want to add, but the core thought work is something you’ve already done if you have a built-out front office and intake team.

Greg Lambert (27:15)
I know this type of customer service agent or intake agent has improved, but I immediately think of the Canadian airline issue, where a chatbot gave false information. What kind of guardrails do you put on this, and as things change, how do you make sure the agent working for you understands those changes?

Michael Anderson (27:43)
Yeah, great question. The thing your agent will be anchored on is the internal process documents you would give to a new hire starting on your front desk or your intake team. That intake team member today probably doesn’t have full case access anyway. So our message is, you don’t have to give it to the AI either.

Michael Anderson (28:12)
You can just give it everything it needs to know to deliver a good client experience, or a good potential-client experience. We are interested in how this might engage with a client portal experience, where clients find out some details on case progress. We have that in our client portal. But today what we were really talking about is that

Michael Anderson (28:40)
you give protocols to a call answering service, you give them to your intake team, so why not also give them to a voice agent and have it do, honestly, a really good job of helping someone get into the system, make progress, and answer good follow-up questions.

Greg Lambert (28:59)
And do you think the process of actually getting the person into the system would, like when you were doing your real estate matter, if you’d had one of these, increase the chances of somebody calling you back?

Michael Anderson (29:13)
Yeah.

Michael Anderson (29:14)
I mean, sometimes I joke that to get a good attorney today, it feels like you’ve got to know a guy, right?

Greg Lambert (29:21)
Debbie from New Jersey.

Michael Anderson (29:22)
Yeah, exactly. You’ve got to know a guy who gets you in, because the systems are really hard. We’re here in California, and the California State Bar did a study, I think a year or two ago, and it’s probably been refreshed since then, that found something like

Michael Anderson (29:42)
one in ten households that want legal services just don’t get them. So when I hear hand-wringing about what work attorneys will do, and we don’t hear this too much anymore, there is a huge ocean of latent demand from people who just don’t get these services and want them. The fact that you can capture more of that by having an AI agent competently

Michael Anderson (30:11)
answer the phone, get them in the system, and enrich their information with things they’ve told the agent, I think that’s a good thing for access to justice. I think it’s a good thing for making the legal system more democratized.

Greg Lambert (30:29)
I’m just curious if that’s one of the benchmarks you’re measuring at Filevine. Do you take a snapshot of what the workload was before a firm put in Filevine and then after, to see what the ROI is? I know this was supposed to be the year of ROI, but I’m still not quite seeing it.

Michael Anderson (30:57)
We know that anecdotally. We know it on both the defense side and the plaintiff side. If you’re representing insurance companies, say, you also want to get through a lot of information, help them manage their legal exposure, and make it predictable. You still have a volume game. And the ability to take on more volume and know

Michael Anderson (31:26)
that where we need to apply expert attention, we still are, and that we’re still delivering a great client experience, that would be a big outcome for us. And frankly, Greg, we’re seeing that. We’re seeing it anecdotally among our customers. That’s exactly what’s happening.

Greg Lambert (31:45)
Now, Filevine is a pretty broad product. It does a lot of things. And in the era of AI, we see a lot of

Greg Lambert (31:57)
very narrowly focused startups coming in that handle one thing really deep, and then you see all-in-one companies that handle a lot of things but maybe shallowly across all of them, good enough in some areas. So, one, how do you view the competition from point-solution companies coming in? And

Greg Lambert (32:27)
how do you make sure you’re not just wide enough, but also deep enough in all of those areas to be beneficial to your customers across all these different services?

Michael Anderson (32:39)
Yeah.

Michael Anderson (32:39)
Great question. A few things on that. One, I love the attention legal technology is getting today. It’s actually really fun. I remember in Filevine’s early seed fundraising, people asked us, gosh, is legal even big enough to build a company? They thought it was so niche. So I think this is fantastic.

Greg Lambert (33:08)
I think they were asking that question a couple of years ago.

Michael Anderson (33:10)
Yeah, right, right. I think it’s fantastic. The legal industry is such a great place for AI applications, so I totally welcome it. I was just talking to one of our partners. We have a whole partner community that does a variety of things that extend our platform. And we talked earlier today about the fact that you can bring

Michael Anderson (33:39)
external connectors into LOIS. Basic things, like an Outlook connector. Of course there are ways to get emails into Filevine and make sure they’re part of the case context, but what if you forget? Or you’re not using Filevine in that specific way? So we announced external connectors. We talked about document management connectors that can be brought in, Outlook,

Michael Anderson (34:08)
and more on the way. And I was talking to one of our partners that does record retrieval, and they said, we’re so excited you announced external connectors. We want our own MCP to be available in LOIS. That’s exactly part of the vision. We’re not going to go deep on every single thing that might help a law firm. So we’ve built great partnerships,

Michael Anderson (34:36)
and we want those to be part of the LOIS platform. There are a lot of problems to solve in this industry, right? The water’s warm, come on in. We welcome you. That’s how we think of it.

Greg Lambert (34:50)
Yeah. One of the issues I’ve had to face at my firm is just the speed at which new products are coming at you, even in the products you already own. How do you help your customers understand what you’ve already released, so they’re taking advantage of it, what’s

Greg Lambert (35:16)
recently released, so they can train on it, and what to expect going forward? How do you handle that communication?

Michael Anderson (35:23)
Great question. I’m going to zoom out, and then I’ll give specific examples. One of the great things about an AI-native interface is that to do the job at hand, you don’t have to know all the capabilities. In the old world, every new capability raised the question, is it in the main navigation? I sometimes joke that if I granted every request from a product manager to put

Michael Anderson (35:52)
their new feature in the main navigation, our customers would need screens that are five feet wide, because everybody wants to be part of the main navigation. Now you don’t need to worry about that. We have an orchestrator that chooses the right tools for the request, for the person asking the question, for the job that needs to be done. So one of the great things we see is that we add

Michael Anderson (36:21)
tools to that orchestrator. It might be a litigation deadline database. People ask questions, and that tool gets used. So one of the cool things is that LOIS will surface the tool for the job whether or not you ever knew it existed. It will call that tool and use it for you without you having to know the right place to click. I just think that’s a cool part of AI-native tooling. We talk about the fact that

Michael Anderson (36:51)
you need to know what you want, but you don’t necessarily need to know where to click. So I would say that. And then of course we do a whole bunch of very traditional things. We have release notes you can find on our site, we do webinars, we do all of that. But it’s amazing how much dissemination happens just from people trying to get a job done, and it ends up calling

Michael Anderson (37:18)
the new capability we just added, and it ends up solving the problem.

Greg Lambert (37:22)
Yeah. I imagine it’s transformed the LEX Summit as well, because this used to be the time when you announced the big rollout for the year, and then you went back and started on the next year. But now it’s weekly, biweekly, monthly releases.

Michael Anderson (37:42)
Totally.

Greg Lambert (37:43)
So when you bring your customer base together for LEX Summit,

Greg Lambert (37:49)
how are you approaching it now that’s different from what you’ve done in the past?

Michael Anderson (37:54)
Great question. You’re spot on. The cadence of the AI era is just faster. The cycles are tighter, and the speed of innovation is relentless. You’ve got to stay on top of it from our perspective, but we also don’t want to overwhelm our customers. So if you notice, the breakout sessions at this conference cover things like

Michael Anderson (38:23)
LOIS for a transactional in-house group, or LOIS for a family law and estate and probate firm doing time and billing. Knowing how to apply it means a lot. And I give credit to our support and expert group. They’re running trainings every single day of

Michael Anderson (38:51)
this three-day conference, because if it doesn’t come alive for you, it doesn’t feel real. But once you’ve had the experience, you’re hooked. If I contrast it with what I was talking about before, a digital transformation launch where everything happens all at once, this is instead a diffusion of dozens or hundreds of small cycles

Michael Anderson (39:20)
that need to happen in terms of onboarding. So it’s still necessary, but once they happen, people get to the other side of it and they get it.

Greg Lambert (39:29)
Well, for the last couple of questions, I want to have you look back a little bit, and then I’m going to ask you to look forward. What’s something that’s true at the 2026 LEX Summit that wasn’t necessarily true at the 2025 LEX Summit?

Michael Anderson (39:45)
Gosh, where to start? We’re here in beautiful San Diego, and there’s not a cloud in the sky. At some of our big Filevine offices, this would be winter, so that’s an obvious one. The diversity of groups that use our platform increases every year. We have litigation founders,

Michael Anderson (40:14)
and the thing that’s interesting is that there’s litigation-relevant tooling in LOIS that brings the whole group together. You could be doing litigation for elder law, or you could be doing litigation representing an insurance company, and you’ve got

Michael Anderson (40:42)
a common language. I think that’s a big emphasis this year. So that’s what I would say.

Greg Lambert (40:49)
All right, now I’m going to ask you to pull out a crystal ball, and we’re going to look to the future. What do you think this industry needs to be prepared for? Maybe a change or a challenge that we may not see on the horizon, but that you’re seeing?

Michael Anderson (41:08)
A question of what’s on the horizon that you might not be anticipating. I’ll go back to where we started, and it’s back to brains. What makes AI do really well at drafting a brief is someone who is really good at drafting a brief. It’s still a force amplifier.

Michael Anderson (41:37)
And I think it’s not just in legal. I’m a father, and I see this with my own middle schooler and high schooler. A lot of schools are now going back to doing things the old-fashioned way to make sure students learn it. There was a wave of homework where you did multiple choice and filled out math online,

Michael Anderson (42:05)
and all of that. And now it’s, no, we actually have to learn these skills first, and then we get amplified. So that’s something I see on the horizon: do it the analog way first, learn it the old way, forge those neural pathways, and then we’ll show you how you can really accelerate it with the new tooling. That’s what

Michael Anderson (42:35)
I think.

Greg Lambert (42:36)
Well, Michael Anderson, Chief Product Officer at Filevine, thank you very much for joining me on The Geek in Review.

Michael Anderson (42:42)
All right. Thanks a lot, Greg.

Greg Lambert (42:44)
And I just want to thank the audience for listening and watching. Please reach out to us on LinkedIn or Substack. Michael, where can people learn more about Filevine and you? Where should they reach out?

Michael Anderson (42:57)
Filevine.com. You can find out a bunch about us there. You can also check out the LEX Summit website and see what we talked about here over the last couple of days.

Greg Lambert (43:08)
Thank you very much.

Michael Anderson (43:09)
All right, thank you.

Last Saturday (October 18th) I was testing some Protégé exports for a project at the firm, and out of old law librarian habit I opened one of the Word files and went to File > Info > Properties > Advanced Properties. Sitting on the Custom tab were five properties I had never seen before.

Lexis is writing these into the Word, Excel, PowerPoint and PDF files that come out of Protégé. The file says it is “AI-Generated,” names “LexisNexis – Protege (Lexis+ AI)” as the provider, sets a human review status of “Unknown – not verified by provider at time of generation,” and adds a disclosure line citing Article 50 of the EU AI Act. Protégé doesn’t mention any of this on screen, and the download dialog has no setting for it.

As far as I can find, LexisNexis has never told its customers about it. (This was validated in my follow up calls with Lexis representatives.) I searched the release notes, the help pages, the terms of use and the May launch announcement for Lexis+ with Protégé, and didn’t find a word about it.

It’s also been going on longer than I realized. After I flagged this inside the firm, one of our researchers sent me a Protégé answer that had been downloaded as a PDF back in August. It has the same five properties, and its AIGenerationTimestamp reads August 20, 2026 at 19:31 UTC (2:31 in the afternoon here in Texas). So Lexis has been stamping files since at least August 20, and I haven’t heard a word about it from Lexis in the two months since.

I sent a detailed write-up to my account team that same afternoon and marked it high importance. I followed up on Tuesday. The answer I have so far, from Tuesday night, is that my notes were shared “with several department leads at LexisNexis for review & counsel on next step actions.” Nothing since. So this weekend I went back in and tested it properly. Last night (Monday October 26th), I talked with a Lexis representative, and someone I know well) and he said that this is something they are working to better communicate with their customers, and they understand that they should have been better at explaining what they are doing, and will work on getting the messaging out. 

What the download looks like from the user’s side

Here’s a normal Protégé answer. I asked it the Texas statute of limitations for breach of a written contract, which is about as plain-vanilla as legal questions get.

Lexis does put a note under every answer: “AI-generated content must be reviewed for accuracy.” I’ll give them credit for that. Click the download icon and you get this:

You can choose the logo, the page numbering, the format and the file name. There is no mention of AI labels anywhere in that box. The Word file that comes out has the five properties. Pick PDF instead and you get the same five, carried over into the PDF’s own document properties, where any PDF metadata viewer will show them. The Zip checkbox just wraps the stamped Word file in a zip.

The test that bothered me most

Protégé has a drafting editor it calls the canvas. I wanted to know whether the stamp tracks what the AI actually wrote, or whether it just gets slapped on anything that leaves the building.

The canvas won’t open without a prompt, so I asked Protégé to create a blank document titled “Human Test Memo” and to leave the body empty. It gave me a page with that title on it and nothing else. Then I turned off Track Changes, deleted everything on the page (including the AI’s three-word title), and typed this myself, one keystroke at a time:

I didn’t use a prompt, a rewrite button or a suggestion along the way. Protégé clearly knew a person had edited the draft, because the card in the chat updated to “Edited on Oct 24, 12:51PM,” and the chat note right under it says track changes “show both AI-generated and manual edits in the canvas.”

Then I downloaded it.

Every word in that file was typed by me, and the file says “AI-Generated” and “This content was generated by an artificial intelligence system.” The PDF says the same thing. (The properties table at the top of the post comes from this exact file.)

I can almost hear the product meeting where this got decided. I’m sure nobody said, “You know what our customers’ human-written memos really need? A hidden certificate saying a robot wrote them.” And yet that’s exactly how the feature behaves right now.

A few other things I found

The “AIGenerationTimestamp” is the download time. I pulled an NDA that Protégé drafted on October 18, downloaded it on October 24, and the file says it was generated on October 24. If anyone ever relies on that field to build a timeline, they’ll get it wrong.

Copy and paste carries nothing. I captured what the copy buttons put on the clipboard, and it’s plain text and simple HTML with no marker of any kind. So the label only follows the people who click Download, and anyone who pastes the answer into Word skips it entirely.

Skills mode gets stamped too. I typed a three-row table of state limitations periods into the prompt myself and asked Protégé’s Skills to give it back to me as an Excel workbook, a Word document and a PowerPoint deck.

All three files carry the same five properties. These downloads come straight from an Amazon storage bucket instead of Lexis’s normal delivery service, so they skip the usual Lexis tracking fields, and the stamp gets written once when the Skill finishes the file. (The Skill also decided on its own to move my citations onto a second sheet as footnote numbers, which is a story for another day.)

Then there’s Review tables, and this one surprised me. I built a review table from a short fictional lease, added an AI column asking for the monthly rent, and downloaded it to Excel with the AI’s reasoning and source citations included. Every cell in that column was written by Protégé. The file has no AI properties and no Lexis tracking fields. I can’t think of a better illustration of the problem: the most AI-heavy file I downloaded all day is the one with no label on it.

For a control, I pulled a case (South Plains Switching v. BNSF, 255 S.W.3d 690) from regular Lexis+ and downloaded it in every format the dialog offers: PDF, Word, WordPerfect, RTF and CaseMap. None of them had any AI properties, including the PDF, which tells me the stamp depends on whether the file came out of Protégé, whatever the format.

In the interest of fairness, there’s one thing I want to head off before someone else raises it. The Protégé files also carry Lexis delivery fields, including one called UserPermID that ties the file to your Lexis user account. Those fields show up on the plain Lexis+ case download too, and Lexis already prints your name in the footer of case downloads. They are not part of this new AI stamp, and I’m not going to pretend they are.

Why Lexis is probably doing this

I get it. Article 50 of the EU AI Act took effect on August 2, 2026, and it requires providers of generative AI to mark their output “in a machine-readable format and detectable as artificially generated.” Systems that were already on the market get until December 2, 2026 to comply with the marking piece. That August 20 file fits the timeline, coming a few weeks after the rule took effect. The rule reaches U.S. vendors whenever their output is used in the EU, and Lexis has no way of knowing which of my downloads might end up attached to an email to Frankfurt. The Commission’s Code of Practice even names metadata as one of the techniques it expects to see.

My read is that Lexis is doing something the EU requires of it and simply skipped the step where customers get told.

Where it goes wrong

The first problem is that nobody at Lexis/RELX told us. This is a change to the files our lawyers send to clients, courts and opposing counsel. A release note listing five property names would have taken someone an afternoon.

Then there’s the way the label gets applied. It rides on the download button and goes onto whatever happens to be in the file. The EU rule itself carves out AI tools that only do “standard editing” or don’t substantially alter the input. A memo a person typed from scratch is a lot further from AI output than that. And the Review tables export shows the label can miss real AI output entirely, depending on which button you click.

The review status bugs me the most, because “Unknown – not verified” never changes. A lawyer can review, edit and sign off on a draft in the canvas, and the downloaded file will still tell anyone who looks that nobody reviewed it, which is a claim about our process that we never made.

And these properties travel. Build your brief from a Protégé download, and the hidden “AI-Generated, not verified” label goes along to whoever receives it. A good metadata scrubbing step will catch it, assuming the document goes through one before it leaves.

How Anthropic handled the same rule

Lexis isn’t the only vendor dealing with this rule. In August, Anthropic announced that text from Claude models launched after August 2 would carry an invisible statistical watermark, built on the SynthID research Google published in 2024. Anthropic also pointed to the EU AI Act, and it chose to apply the watermark worldwide. The reaction was loud. Dozens of users on X posted screenshots of their canceled subscriptions. John Gruber called the word-choice nudging behind it “a perversion of writing.” Others worried that a lawyer who only used Claude to tighten up a brief could have the whole thing flagged as AI-written, which is pretty much what happened to my hand-typed memo in the Protégé canvas. Anthropic told reporters it saw no measurable rise in cancellations and stuck with the plan.

The two approaches work differently. Anthropic’s mark lives in the word choices themselves, so it survives a copy and paste, and Anthropic says it will offer an API that checks text for it. Lexis’s stamp is five lines of metadata that disappear the moment someone copies the text out or runs the file through a scrubber. The bigger difference for me is disclosure. Anthropic announced the watermark, published a technical explanation of how it works, and took the criticism in public. Lexis has been stamping files since at least August 20, and its customers are finding out by opening the document properties themselves.

What I’d like to see from Lexis

I’d start with a public release note that names the properties and says when they started. I’d also like the stamp to use the canvas’s own record of AI versus manual edits, since the canvas clearly already knows the difference, and I’d like the review status to update when a named user signs off. If there’s a reason Lexis can’t give firms with no EU exposure an admin setting for this, I’d like to hear what it is.

In the meantime, firms should check whether their metadata cleaning tools catch custom document properties, and should start asking every AI vendor they license whether exports are being stamped. I’d be surprised if Lexis is the only one with a December 2 deadline on the calendar.

So here’s my question for my fellow innovation and KM folks: if one of your AI vendors started writing statements about your firm’s work product into the files you send to clients, how long would it take your firm to notice? It took me two months to find out that Lexis did it.

Greg Lambert and Marlene Gebauer take The Geek in Review on the road to Atlanta for a conversation following the GDS AI Summit. Meeting with leaders from across industries offers a fresh perspective on familiar legal technology challenges: messy data, uncertain ownership, and the pressure to turn artificial intelligence experiments into measurable business improvements. Their shared takeaway offers some reassurance for legal innovation teams. Law firms face many of the same obstacles as their clients, and cross-industry conversations provide useful lessons for moving forward.

Data quality anchors the discussion, with Greg offering a memorable description of law firm information: “Our data puddles are mud puddles.” Greg and Marlene examine structured data, unstructured data, and the less obvious problem of missing or unidentified information. Some organizational knowledge lives exclusively in the heads of experienced professionals, creating continuity risks when those people leave. The conversation also explores how priorities differ across industries, from protecting confidential information in legal services to safeguarding reputation when AI-generated content reaches an audience.

Hidden workarounds emerge as another barrier to meaningful process improvement. Drawing on a conference discussion about nurses solving problems outside formal reporting channels, the hosts consider similar habits among legal assistants, paralegals, attorneys, and technology teams. When employees quietly resolve recurring problems, leadership receives an incomplete picture of daily operations. Over time, the workaround becomes the process. Introducing AI without understanding those informal practices risks embedding existing problems into a new system.

Greg and Marlene also discuss the discipline required to run useful AI pilots. Their breakout group narrowed a broad data challenge into a more focused project around client data intake, ownership, and tracking. The exercise prompted questions about pilot duration, accountability, and measurement, including the risk of an extended experiment becoming a permanent process without a deliberate decision. Adoption numbers offer only part of the picture. Understanding whether people use a tool effectively requires clearer objectives, meaningful metrics, and responsibility for gathering evidence.

The episode closes with a larger strategic question: Should AI adapt to existing business operations, or should organizations rethink their operations around AI’s capabilities? Greg questions the value of adding AI to established products while preserving the same underlying approach, while Marlene distinguishes enterprise-wide integration from solutions serving specific legal workflows. Alongside reflections on their own conference presentation and an evening at the Margaret Mitchell House, the hosts bring the conversation back to practical priorities for law firm leaders: define the problem, understand the process, and evaluate results before expanding an AI initiative.

GDS Group and AI Summit Events

Listen on mobile platforms: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Apple Podcasts⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ | ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Spotify⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ | ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠YouTube⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ | ⁠Substack

Email: geekinreviewpodcast@gmail.com

Music: Jerry David DeCicca

Transcript:

Greg Lambert (00:00)
In law, our data puddles are mud puddles. That’s how dirty our data is.

Marlene Gebauer (00:10)
Welcome to The Geek in Review on the Road, the podcast focused on innovative and creative ideas in the legal profession. I’m Marlene Gebauer.

Greg Lambert (00:17)
And I’m Greg Lambert, and we are in Atlanta, Georgia, this week at the GDS Summit on AI.

Marlene Gebauer (00:26)
Yeah.

Greg Lambert (00:27)
We just wrapped it up, so we thought we would take a few minutes and walk through what we just experienced.

Marlene Gebauer (00:34)
Yeah. So this was a new conference for both of us. I think it was very helpful because we were meeting a ton of new people, industry leaders in different companies as well as builders. You had the gamut of people who were involved in the AI network. We got lots of insights, were able to network with a lot of people, and learn about a lot of new products.

Greg Lambert (01:01)
Yeah, and I think the way that we sold this is we’re around a bunch of the clients, the potential clients, which was very true. The way the conference was set up was that it was split into, I think, three different things. We had presentations that people were giving onstage. We had a breakout session, and there were four tracks. We both picked the same track because we both deal with data issues, so we did that. Then there were one-on-one talks with the vendors, where we’d spend about 30 minutes with a vendor. I think I did five or...

Marlene Gebauer (01:45)
Yeah, I think I just...

Greg Lambert (01:46)
Four of them. Yeah. It’s a well-done conference. They’re kind of fun. This is actually my second one. I did one in Memphis earlier this year, and I...

Marlene Gebauer (01:56)
[Unclear phrase about Greg returning to the conference.]

Greg Lambert (01:57)
Did it. We’re at a really nice hotel, The Starling in Atlanta, which is just down the road from my old firm, King & Spalding. So I get to walk by the old place.

Marlene Gebauer (02:10)
Yeah, and part of it was that we were doing a presentation about hard questions about AI. We talked a lot about data and the necessity for good data in order to be within the new world of artificial intelligence and make it work appropriately.

Greg Lambert (02:29)
A lot of it was that we’re dealing with an issue that’s not a new problem. It’s a systemic problem over many years. But it was interesting because there were probably about 15 or 20 people in our group.

Marlene Gebauer (02:48)
Yeah, from all different environments and worlds. The cool thing about it is that the problems were all the same. It surprised me because they have customers, we have customers, and the same sort of concerns.

Greg Lambert (03:07)
Yeah. I think a lot of it was comforting in that we’re not as far behind as we thought we were. We’re probably ahead in some places. I think it was a good balance. It was good to hear from people outside the legal industry on what kind of issues they’re facing with their data.

Marlene Gebauer (03:26)
I always like that because you’re kind of outside of your comfort zone. It’s good to hear what your customers and potential customers are also struggling with. Some of the things we covered were the investment proposition and how to pick use cases that won’t [unclear phrase about pilots], visionary architecture that works at the enterprise level.

We had a debate today, a friendly fireside debate. They were talking about whether AI should change the enterprise or whether the enterprise should change AI to what it needs. We had a kind of point-counterpoint on that. So that was fun.

Greg Lambert (04:16)
Then last night we went to the Margaret Mitchell House, where she wrote Gone with the Wind. That was interesting because, in addition to the house, they had a little side place that they turned into a disco. It was kind of wild.

Marlene Gebauer (04:33)
We had the calm side with the rocking chairs, and we had the disco side. So you can guess where we went.

Greg Lambert (04:39)
Yeah, that was definitely on the rocking chairs. If you had to point out one or two things you took away from this, what’s something that stood out for you?

Marlene Gebauer (04:50)
Something that stood out was, again, the commonality that everybody has messy data and is struggling with that and how to solve for that. Although it was interesting, we did have one view that you can put unstructured data into solutions and it will work fine. I don’t think any of us are quite buying that. But hey, man, if you have something like that and you want to talk to me about it, let me know.

Greg Lambert (05:21)
We did talk a lot about things like data lakes, data [unclear term], and today was data puddles. My comment was that, in law, our data puddles are mud puddles. That’s how dirty our data is. There was an interesting thing that I think we don’t think about in law directly because we think so much about what we have that we need to protect.

Because it’s...

Marlene Gebauer (05:51)
Security was a big thing. I mean, security is a big thing for everybody, but particularly for us.

Greg Lambert (05:56)
The counterpoint to that was, I talked to one guy, Joe, and there was a woman from Condé Nast, Nan Lee [name uncertain], who I did not get a chance to...

Marlene Gebauer (06:08)
Who ran one of the sessions.

Greg Lambert (06:09)
Talk to. The way Joe had put it is, most of us are protecting our data. That’s our secret sauce, and it’s highly controlled. In entertainment, with [organization name unclear] and Condé Nast, they talk about how they give away their secret sauce. That is the value they have. Their issues around AI are much more around reputation.

Marlene Gebauer (06:45)
There’s a representation out there that isn’t correct. Yeah. There’s a lot of...

Greg Lambert (06:50)
It looks like they’re putting content out that’s AI...

Marlene Gebauer (06:53)
AI-generated. There’s a lot of backlash on that.

Greg Lambert (06:55)
Yeah. It was interesting to see the flip side of AI, where it’s highly used, but they have to be very careful about what it is. It’s not the data they’re concerned about; it’s the reputation. That was an interesting point of view.

Marlene Gebauer (07:15)
I liked that, in our group, we were talking about three issues: structured data, unstructured data, and no data. We all had to break out into groups and come up with a solution proposition to a particular issue that we determined. We didn’t pick the no-data issue, but no data could be a lot of different things. No data could be that you really don’t have any data, or you don’t know that you have any data, or you don’t know what the data is.

Greg Lambert (07:48)
Yeah. It’s also the data that’s locked in between the ears of your subject matter experts and professionals. That single point of failure was raised a lot.

Marlene Gebauer (08:00)
[Unclear phrase about how often this arises in their experience.]

Greg Lambert (08:04)
Yeah, you get the person who’s been there 30 years and knows how to do something. I did think the funniest thing was one guy saying his HR department won’t let him say they got hit by a bus. Now he says they won the lotto. That’s a nicer way to put it.

Marlene Gebauer (08:20)
Which is nice.

Another thing was how people were coming up with solutions. This is all public knowledge, but nurses will work around problems to solve them. I don’t think that’s just nurses; I think that’s everybody. We all try to do workarounds to get the job done that we need to get done, versus, okay, let’s take this and dig in and make the change.

Greg Lambert (08:50)
It also gives you incomplete information. When you’re trying to solve for an issue, if 97 percent of that issue has been worked around and you only know the 3 percent that’s going to be reported, you’re dealing with incomplete data. I thought that was a good point. I did like the fact that the guy who brought up the study said his wife was a nurse, and she punched him in the arm when he was talking about this.

But I think we see the same things, whether it’s legal assistants, paralegals, associates, or IT. A lot of what we run into when it comes to process are typically things we put up with, work around, and build into the process. But again...

Marlene Gebauer (09:46)
The workaround becomes the process, which is the danger.

Greg Lambert (09:50)
I think that’s something that has to be taken into consideration when you’re looking at addressing that issue. You’re looking at process improvement, and you don’t know what the whole process is. You can create a solution that builds in the very problem you’ve been working around.

One of the other good takeaways from our breakout session was that we essentially had to come up with a plan for how to address it. I think we were told on day one that our problem was too big. It’s not surprising. We narrowed it. The problem we came up with was intake of client data: how you do that, who has ownership of it, and how you track it.

One of the things listed was how long this should take. We decided we were going to take the pilot, do it in one department, and work through the process. But the question came up: How long do you do this pilot? It was 30, 60, or 90 days. My pick was 60.

Marlene Gebauer (11:15)
Yeah, we had a wide variety of responses in there.

Greg Lambert (11:18)
Yeah. Another person was like, 90, so we split the difference and did 60 to 90. But one of the things brought up was that the longer these pilots go, the more of a chance they stop being pilots and become the process.

Marlene Gebauer (11:38)
People forget you’re in a pilot stage. It’s sort of like, “Okay, what happened to my access?” or whatever.

Greg Lambert (11:44)
Yeah. That was something you keep in the back of your mind as you’re doing this, but I think it was nice to have it specifically listed out as something you have to do. It was more encouraging that these pilots be much shorter, that there be almost no meetings set up on them, and that you have metrics at the beginning and metrics at the end. We did talk about how sometimes you don’t have good metrics.

Marlene Gebauer (12:20)
What are good metrics, and actionable metrics? That was also an area of discussion because oftentimes either the product itself or the vendors themselves don’t have that, and you don’t have a way to track it. Or it’s just not being tracked.

Greg Lambert (12:46)
Yeah. A lot of that’s difficult to do because there’s a lot of stuff where we have a feeling of how things are, but feelings are really hard to measure.

Marlene Gebauer (13:00)
About adoption, right? People are using it, but how do we know they’re using it well? How do we figure that out? That’s a harder metric to figure out.

Greg Lambert (13:11)
Yeah. The whole deal with ownership and reporting, not having as many meetings, and having metrics on the back end. It’s things you know you need to do, but it was nice to have it laid out.

Marlene Gebauer (13:29)
Yeah, and people may have to do it in a way where they take responsibility to develop something that allows them to get the information they need.

Greg Lambert (13:42)
Yeah. I was laughing on the first day because the way these are set up is you spend the first couple of days identifying the problem. It was interesting because I was identifying a problem, and one of the guys was like, “Well, here’s what we did,” and I was like, “Well...”

Marlene Gebauer (14:04)
No, no [unclear].

Greg Lambert (14:05)
Yeah, no solutions yet. We’ll get to that. So it was nice.

It was over three days, but it was a half day, a full day, and a half day. It was really two days of meeting. Having it where the first three-quarters was identifying problems that exist across different industries, and then coming up with solutions, but really more about process, I thought it was a good setup.

Our master of ceremonies, Tim Cassidy, was an excellent host. He invited us for the last day. We had a 30-minute session. It was obviously the best session of the entire 30 minutes that we were there.

Marlene Gebauer (14:58)
Standing ovation. Standing room only. All of that.

Greg Lambert (15:02)
Absolutely. We were one of the last ones.

Marlene Gebauer (15:06)
Which is...

Greg Lambert (15:08)
We talked a lot about the podcast. We talked about where we are in the legal industry. It was interesting because we had some people who were former attorneys and people who were legal-adjacent, quickly identifying some of the issues.

Marlene Gebauer (15:27)
People came up later. It’s like...

Greg Lambert (15:28)
Yeah. “We have those same issues,” or, “I deal with these issues with our in-house teams, general counsel at our work.” I thought it was fun to talk about what we’ve seen in the legal industry and see the reaction from other industries. They were able to easily relate to some of the same issues we were running into.

One of the things we said in our session contributed to the next session when they had the debate. Again, do you set up AI to influence the way the business is operating, or does the way the business operates influence AI?

We mentioned the situation where we thought, or at least I thought, there were a number of legal vendors that slapped AI on top of their existing product to make it more efficient at doing what it always did. There were others that came in as startups and created new products for the industry. I thought slapping AI on top of the product to make it much more effective through AI, but still doing the same thing, was not a wise choice.

We saw that in the debate because they were saying, well, do you slap AI on top of your business and make your business do what it does better with AI? Or do you look at it and come up with a new...

Marlene Gebauer (17:17)
Process based on what AI could do. I think the panel’s distinction was that they were looking at AI integration at an enterprise level, within all aspects of the enterprise. I think we were talking a little more about solutions dealing with our business processes, but also impacting our clients, and more attorney- or practice-facing types of...

Greg Lambert (17:46)
Situations. Atlanta was nice. We...

Marlene Gebauer (17:49)
I enjoyed getting out of the 90-degree heat and humidity.

Greg Lambert (17:53)
Atlanta’s always a fun town. I think it was enjoyable. Thank GDS for inviting us out and having us talk. I enjoyed it, and I’d do it again.

Marlene Gebauer (18:06)
Yeah, absolutely. Okay, bye, everybody.

This week we talk with Ross Guberman, founder and CEO of BriefCatch, and Heather Rodriguez, the company’s enterprise account executive. Ross built BriefCatch to put the lessons from his books and workshops in front of lawyers while they write. It started as a rules engine that flagged everything from typos to sentence flow. He admits he was never a coder, and says he would have looked for a technical co-founder if he had known the term existed. Today BriefCatch pairs that original engine with generative AI to review whole briefs, motions and judicial opinions for persuasiveness and consistency. The company has spent much of 2026 on checking how accurately briefs describe cases and statutes.

Greg opens by testing the premise that AI makes a first draft cheap and abundant. Ross agrees that ChatGPT, Claude or Gemini will produce a serviceable draft of an NDA, a motion or a client letter. He warns that the polish can be “dangerously seductive.” He expects the gap between weaker and stronger legal writers to widen, and thinks the best writers will produce better work than the profession has ever seen. Heather names judgment as the scarce skill: can the lawyer tell whether a polished draft is accurate and strategically sound? Marlene asks how junior lawyers will learn when they start by editing. Heather points out that lawyers have always started from a brief bank or a prior motion, so strong editing has always been part of strong writing. Ross suggests running a draft through a second model and asking it to verify and source every legal and factual statement.

That leads to RealityCheck, BriefCatch’s citation verification tool. Heather explains that it checks whether the cited opinion supports the proposition, pulls up the relevant language and flags potential problems. An experienced lawyer can disagree with its reading, and the lawyer still owns the argument. Ross adds that running RealityCheck on briefs from the 1990s through the 2010s turns up plenty of case law errors, so the problem predates generative AI. The talk then turns to how alike AI prose sounds, and Marlene warns she may lose her mind if she reads “it hits differently” one more time. Heather describes how BriefCatch offers choices the attorney can accept or reject. Ross gives the example of lawyers who use “indicate” for everything; the tool suggests more precise verbs such as specify, convey or reflect. On Anthropic’s plan to watermark text written by Claude, Ross praises the accuracy of the AI detector Pangram. He hopes the profession will start asking a better question about AI-assisted work: was it a good brief?

Heather then describes where enterprise adoption stands. The Legal Technology Hub map has grown so crowded you need a magnifying glass to read it. Firms now ask whether a product is the best fit for a specific use case and whether it can replace tools they already own. Workflow fit, ROI and security reviews dominate the conversation, and she has seen firms that love a product spend months in security and procurement. Ross credits his reputation in legal writing with helping BriefCatch earn trust, though the company still answers the same questionnaires as every vendor. He walks through the WordRake acquisition, closed in January, which brought Gary Kinder’s conciseness and plain language technology and a new customer base to BriefCatch, and he hints that more acquisitions may follow. BriefCatch is also expanding past its Word plug-in into web apps, Outlook, an admin platform and PDF reports for courts, while Ross notes that most lawyers still do their real work in Word. When Greg asks about an MCP integration with Claude or OpenAI, Ross says one could arrive before the end of the year.

Marlene asks what is true today that wasn’t a year ago. Heather says AI has become a normal part of legal work, and hands-on experience has taught lawyers what the tools do well and where they fall short. Ross calls the weekly stream of new possibilities energizing and overwhelming, for founders and customers alike. Looking ahead, Heather expects firms to build their own AI ecosystems and pick the right tool for each stage of the work. Ross predicts that courts, which are quietly experimenting with generative AI, will use it far more within a year. Once courts signal their approval, he expects lawyers to stop debating whether to use these tools and start choosing which ones. Marlene drops off near the end with technical issues, so Greg wraps up the conversation with both guests.

Resources mentioned

• BriefCatch
• RealityCheck (BriefCatch support page)
• BriefCatch launches unified legal writing suite (LawNext, Sept. 2026)
• BriefCatch acquires WordRake and its 12 editing patents (LawNext)
• WordRake
• Pangram AI detection
• Anthropic to embed invisible watermarks in Claude-generated text (VKTR)
• Study finds AI-generated prose in more than 50 federal appellate opinions
• Legal Technology Hub
• Ross Guberman bio (Legal Writing Pro)

Transcript

Marlene Gebauer (00:07)
Welcome to The Geek in Review, the podcast focused on innovative and creative ideas in the legal industry. I’m Marlene Gebauer.

Greg Lambert (00:14)
And I’m Greg Lambert. And this week we are joined by Ross Guberman, who’s the founder and CEO of BriefCatch, and Heather Rodriguez, who is the enterprise account executive at BriefCatch. Ross, we should have had you on a long, long time ago. But we’re finally glad to get you on here.

Ross Guberman (00:34)
Thanks, I appreciate it.

Heather Rodriguez (00:35)
Thanks for having us.

Marlene Gebauer (00:37)
So for years, BriefCatch has focused on helping lawyers become better writers. But legal writing is changing quickly. Generative AI is producing more of a first draft. BriefCatch has acquired a long-time competitor WordRake, and the company has expanded into verification, professional development, and enterprise writing tools.

Greg Lambert (00:57)
And I guess that raises a bigger question than just whether AI writes well. It’s, you know, what happens to lawyers when generating the words becomes really kind of the easy part. So again, guys, welcome to the show. Ross, I wanted to kick this off by letting you have a chance to kind of give the listeners an overview of what BriefCatch is and what it’s become over the past year or so.

Ross Guberman (01:25)
Yeah, absolutely. So the original idea behind BriefCatch was to take all my books and my tips and my workshops and try to somehow get all that into a tech format that would make it easy for busy lawyers to access those kinds of edits and tips and suggestions while they’re actually working. So the original product was, as you can guess, pure code. It was kind of an editing rules engine that just grew and grew over the years and we would tackle everything from you know very specific kinds of typos or errors that lawyers make to opportunities to have better flow, more variety in sentences and everything in between.

So that kept us very busy and very excited for several years. But again, we stuck to our format of a rules engine and spotting patterns in language, trying to match up all the teachings to an actual coded rule. And then also we got a lot of input from judges and partners and firms for more and more ways we could help lawyers write more effectively and edit more effectively. Now, of course, as everyone knows, we’re in the Gen AI era, which is a little scary on the pedagogical front, as you already alluded to, but also really exciting on the tech front. So now we’re entering the realm of verifying accuracy and integrity in discussions of cases and statutes.

So that’s been occupying us quite a bit this year. And we’re actually moving into something that’s really exciting for me personally, which is a much broader concept of legal writing improvement and legal editing. So with Gen AI and our own proprietary tools and our original rules engine, we’re now going to be looking at, for example, an entire brief or motion or judicial opinion and we’re able to spot opportunities for better persuasiveness, more consistency, like all the kinds of things that Gen AI enables you to do quite well, as long as you have a lot of guardrails in place. So I’ve kind of lived through this whole legal tech evolution myself, going from like pure code to Gen AI.

So it’s been really exciting. And I’d say right now is probably the most stimulating time since I started.

Greg Lambert (03:38)
Now, were you a coder when you started or were you like me, you just kinda jumped in and you knew this could be done?

Ross Guberman (03:46)
I absolutely was not and I have to say I’m still not really a coder. I picked up a little bit by osmosis here and there. So you know, I didn’t really know that I’d be a founder of a tech company, to be honest. If I had to do it over again I would have gotten a technical co-founder. I didn’t even know that term, by the way, existed at the time. But I’ve managed pretty well. You know, I play to my strengths, I know my weak spots, so we hire accordingly. But no, I did not know how to code. And I wish I had. I mean I really admire the skill.

I think it’s an important skill even today when you can do so much code by automation. But no I’ve had to do it mostly as I said by osmosis.

Greg Lambert (04:29)
I imagine that the LLMs helping with the coding have also been beneficial as well.

Ross Guberman (04:38)
Yeah, I mean, you know, people like video games and I do too, but I really, really like watching, like if you put code up and you ask ChatGPT to optimize it, and you kinda watch and, you know, it’s mesmerizing in a very nerdy way, and you can get a little bit of a sense of how to improve code and automate the process that way too.

Greg Lambert (04:59)
Well, I wanted to start off by asking you something. I think we all think that AI is making writing cheaper. That we’re able to actually, you know, produce at least a draft much more quickly, and reduce the effort that it takes to make the draft. And so I’m gonna ask you a question, but I want you to first start with: Is the premise of this question correct? And that is if the producing of a first draft becomes cheap and abundant, what becomes scarce and valuable in legal writing? And I really wanna say, does AI make a first draft cheap and abundant?

Ross Guberman (05:47)
Well, there’s no doubt that even today, you can go to ChatGPT or Claude or Gemini and ask it to generate a first draft of, let’s say, an NDA, you know, a motion in limine, even a letter to a client, give it some background, give it the jurisdiction, give it some of your own perspective. And you will get, you absolutely will get in many ways a serviceable first draft. Now it will be dangerously seductive in the sense that it will often look much more polished than it really is. But it is, as you said, cheap and abundant, right? And I think, you know, people want to deny the truth because they’re worried about overreliance.

And I get it, but we should just admit that although you can find some counterexamples, in general the first draft you’re going to get is decent and it will hit most of the hot spots. So I think the premise is correct. It’s correct with the caveat that it’s often dangerously seductive in its exhaustiveness and polish. And that kind of leads me to the answer to the actual question, which is then that’s what’s going to become really, really scarce. Because you’re gonna end up having a very broad and ever-widening gap between the weaker players and the stronger players in legal writing. And because of the growing temptation to rely on these tools, not just for the first draft, but in let’s say kind of the middle stages, what’s gonna become really rare, I would say, will be a truly top-notch, well-thought-through, tailored product.

I do think the very strongest players, like the people who are really gonna shine in whether it’s contract drafting or litigation or appellate advocacy, they will probably be better writers than we’ve ever seen in the history of legal writing because they will be able to harness all of their own knowledge, their own perspective, and learn how to use the power of these tools effectively. And that is actually, you know, pretty exciting, right? The ceiling’s probably gonna be higher than ever.

Greg Lambert (07:47)
And Heather, are you seeing the same thing?

Heather Rodriguez (07:50)
Yeah, well in my opinion, I think that what’s going to become more valuable is the judgment. So yeah, AI can make a plausible first draft and it can be polished, but does it necessarily mean that it’s accurate and is it strategically sound? So the interesting part is yeah, it may become easier to produce something that looks good, but how do we know if it’s actually good? And so I think it’ll be incredibly valuable for attorneys to be able to evaluate and verify and actually prove that the work is accurate, you know, and a great work product.

Ross Guberman (08:26)
Heather talks to firms every day and I think I mean you can speak to this much better than I can, Heather, but there is some anxiety, right, about especially junior lawyers and their skill sets.

Marlene Gebauer (08:38)
Yeah, actually I wanted to focus on that, focus on like lawyer skills and how we’re seeing that change and how we’re seeing a change in the writing process. So Heather, I mean, young lawyers have traditionally learned through, you know, doing drafts, getting feedback, rewriting,

Heather Rodriguez (08:57)
Okay.

Marlene Gebauer (08:58)
But AI now of course produces this starting point, you know, this basic draft. So if lawyers are learning by editing rather than creating the work from a blank page, what are the types of skills that they’re going to need in order to do the writing process?

Heather Rodriguez (09:19)
So I look at this a little differently. I don’t think lawyers have ever really started from a complete blank page. There’s always a brief bank, a precedent, prior motion, or something that another lawyer at the firm has written that provides that starting point. So I feel that part of becoming a good lawyer has always been knowing how to take that starting point, think critically about it, make it better for your particular argument or client. And although AI makes finding or creating that starting point dramatically more efficient, you then still edit it, but I don’t think being a great editor is a lower level skill at all.

In fact, with AI, I think it’s becoming one of the more important skills that a lawyer can develop. You have to have the ability to look at something and ask, is this a strong argument? Is it persuasive? Does this belong here? So I don’t think that the concern should be that we’re training lawyers to become editors. We should have always been focused on training them to become better writers. And in my opinion, great editing is part of great writing.

Marlene Gebauer (10:27)
Yeah, so basically critical analysis isn’t gone. That still continues.

Heather Rodriguez (10:32)
No. Correct.

Greg Lambert (10:35)
Yeah, well, I guess I want to push on that one just a little bit because you hear a lot about, you know, because the AI looks so good right out of the gate, that the lawyers, especially those who are working in an area where they may not be all that knowledgeable in the ins and outs of it are relying too heavily on it, giving a little bit of too much cognitive offloading to the AI. When either of you are talking to lawyers, are you finding that to be the case? And if so, how are more experienced lawyers kind of pushing back and making sure that the younger lawyers aren’t kind of offloading?

Is that through the red pen or how are they doing that?

Ross Guberman (11:26)
Well there are you know, there are some sophisticated uses of AI that you’re seeing right now in firms and elsewhere, at all parts of the food chain, like senior partners down to first year associates. So for example, one thing that I’ve recommended and many people have tried this and found it to be successful, is if you are gonna use the Gen AI for the first draft, and let’s be honest, a lot of people are going to, regardless of what any of us on this podcast think they should do. And you immediately switch to another model and you have it scrutinize the draft and what I recommend is have it verify every single legal and factual statement and every single statement related to case law.

Verify that it’s accurate, but give you a link or some sort of source. You can actually already capture a lot of the functions of a human editor yourself, right? But you have to, again, know how to add that level of scrutiny. That’s not gonna happen automatically. Like that’s the kind of thing that people are realizing they can do so that they’re able to actually maximize the value of these tools.

Greg Lambert (12:37)
Yeah, well that brings me to one of the tools that I think that you’ve developed in order to kind of help with this as well, and that’s RealityCheck, which can verify that you know cases exist and kind of help to make sure that the AI is sticking to the fact patterns and the reality and isn’t going off and inventing something. So talk to us about how RealityCheck works, what the effect is, and then kind of how the lawyers interact with that when it works and when they disagree with it.

Ross Guberman (13:18)
I’m gonna invite Heather to describe RealityCheck because she deals with it every day and I’m a little bit biased since I created it, but I will give you one quick preview and that is if we use RealityCheck on briefs written in the 1990s or the 2000s or the 2010s, we uncover all sorts of errors in relation to case law. That’s right. Yeah, I know.

Greg Lambert (13:38)
No, sorry, it only started after 2022.

Heather Rodriguez (13:41)
Mm-hmm.

Ross Guberman (13:43)
There’s a lot of romanticizing of the past going on. So anyway, with that little preview, I’ll turn it over to Heather to tell everyone about

Heather Rodriguez (13:49)
[Laughter]

Ross Guberman (13:52)
RealityCheck. That’s RealityCheck, one word, by the way.

Heather Rodriguez (13:55)
RealityCheck is there to inform lawyers’ judgment. It’s not there to replace it. So the big difference here is between confirming that a case exists and analyzing whether that case actually supports the proposition being made. So that’s why we created it. RealityCheck can surface the relevant language, explain what the opinion says, identify any potential issues, and give lawyers a much more efficient way to evaluate that citation. So we’re not only confirming the citation exists. But legal arguments can be nuanced. So reasonable lawyers can interpret authority differently. So maybe an experienced lawyer may disagree with RealityCheck, and the goal isn’t for the technology to have the final word.

Again, we want to give the lawyer another layer of analysis to consider and the information that they need to make their own determination. So, again, the lawyer always needs to own the argument and the final work product. So, I think that’s where legal judgment should remain. We’re there to provide that assistance.

Marlene Gebauer (14:54)
So one criticism of AI-generated writing is that everyone’s writing begins to sound increasingly similar. Like if I hear “it hits differently” one more time,

Ross Guberman (15:07)
Mm-hmm.

Marlene Gebauer (15:07)
I think I’m going to go insane.

Greg Lambert (15:09)
How did that land? That’s what I wanna know.

Ross Guberman (15:10)
And how does it land? Also,

Heather Rodriguez (15:10)
Yeah.

Ross Guberman (15:12)
Something is load-bearing. Every point is load-bearing.

Greg Lambert (15:14)
Yeah.

Marlene Gebauer (15:16)
So we all understand.

Greg Lambert (15:18)
Yeah.

Marlene Gebauer (15:19)
But as AI becomes embedded in legal writing, how do we preserve lawyers’ individual voice and style so that they don’t sound like everybody else?

Ross Guberman (15:29)
I’m gonna toss that over to Heather too, because she gets this question a lot and I think she has a good answer.

Heather Rodriguez (15:34)
And there’s a real risk in that. And you don’t want everyone to sound the same, but that’s not how we think about it here at BriefCatch. We’re here to preserve the attorney’s voice and actually help them strengthen their writing. And so we’re giving attorneys the choice and examples for inspiration. Rather than saying, Here’s how you should write this, the attorney is still making that decision. So they can accept a suggestion, they can completely reject it entirely. So if there’s, let’s say, a specific phrase or stylistic choice that feels authentic to them, then they can keep it. So we’re not trying to standardize everyone’s voice here.

We’re helping firms standardize the quality of the writing that goes out the door while allowing individual lawyers to still sound like themselves. So it’s just not about producing a better document. By leveraging BriefCatch, they’re able to learn those different choices and that improves their own voice as a writer over time.

Ross Guberman (16:26)
Well a good example is a lot of lawyers and judges will gravitate toward indicate, whatever they’re talking about. The record, the docket, the case, the client, the witness, opposing counsel, indicate, indicate, indicate, indicate, indicate all day long. So one of the first challenges I took on was to try to give people a menu of more specific options when they write indicate in the first draft, like specify or convey or reflect or state or say. So that’s a good example of what Heather is getting at. So in regular human produced writing, there’s a lot of homogeneity too, because we’re all inclined to pick the most obvious verb or noun or term of art.

So a lot of what we do is actually again try to make people sound more like themselves and a little bit more specific and more precise as an antidote to this very real problem of, homogenization.

Greg Lambert (17:21)
Ross, there was a recent announcement by Anthropic and I think other companies are gonna do this as well, to where they’re gonna start adding kind of unseen watermarks in the AI writing, which you know if you think about somebody that’s using it to create something whole cloth and using it, yes. But you know, a lot of people were arguing that, you know, they’re using the AI to like fine tune some of the arguments that they’re having and yet they’re worried that this you know it’s going to be able to look like it’s been created completely by AI. Is that a worry that you’re hearing from your customers?

Ross Guberman (18:04)
Yeah, I mean I think a lot of people were surprised by that announcement from Anthropic. I know that I personally and many others have used a tool called Pangram that is, to be honest, incredibly accurate in predicting the percent of text that was generated by AI. I’ve tried to trick it and I’ve failed.

Greg Lambert (18:25)
I can’t trick it either.

Ross Guberman (18:26)
You can’t trick it. I mean it’s an incredible tool. I like to give praise where praise is due. What I would say, and this is especially true in our particular profession, is I think we need to get over the sort of automatic gotcha and smugness in catching someone using AI. You might have heard the I think it was an academic study a few weeks ago. It looked at a bunch of federal court opinions and singled out, I don’t know, a handful, you know, accusing them of using AI to write the opinion. But no one ever gets to the next question, which is: Was it a good brief?

Was it a good opinion? Was it a good LinkedIn post? Right? There are plenty of human-generated examples of all the above that are crappy. There are some AI-generated examples of all the above that are also not very good, but there are a lot that are really great as well. Look, there’s a reason there’s a market for the watermark. There’s a market for Pangram. There’s a reason we have this database of hallucinated case studies all over the country. But what I would guess is that we’re gonna probably get to a point pretty soon where there’s a little bit of nuance and it’s not treated as a crime.

Like you’re not automatically presumed guilty here unless the AI-generated content itself is otherwise flawed.

Greg Lambert (19:46)
Makes sense. Heather, I’m gonna change gears a little bit on you and talk about adoption of the technologies within law firms. I know when you guys go into law firms and you talk about things like enterprise adoption of your software, what are you seeing in law firms now? Because I know I’m being asked a lot about what’s the return on investment that we’re getting. And in fact I’m seeing some changes in the way law firms are structuring their AI teams now where IT is now stepping in and doing a lot more. And so what are you seeing as far as behavior, changing the lawyers’ behavior, the security issues, proving ROI, what’s the conversation that you’re having these days with law firms?

Heather Rodriguez (20:46)
There’s so many pieces to it and the conversation has definitely changed. There’s so much more technology today than ever before. I remember two years ago looking at that Legal Technology Hub map of all the different tech and today you can’t even read the names.

Ross Guberman (21:02)
Yeah.

Marlene Gebauer (21:02)
So small.

Heather Rodriguez (21:03)
It’s so small. You need a magnifying glass. And so I think firms are almost overwhelmed with all the options. And they’re not just asking, does technology work? We know it works. Is this the best technology for this use case? Can it help us consolidate some of the tools we already have? When it comes to workflow integration, I think that’s still a big factor. Most AmLaw firms, they have a significant tech stack. So they’re thinking about how everything works together for the attorney rather than just adding another tool for the sake of having it. They’re being much more intentional about the tools that they’re choosing. And I think changing lawyer behavior is less of an obstacle than people assume.

Most firms that I speak with, they are wanting to be the most innovative. And I think the big challenge, though, is giving attorneys the technology that fits naturally into their workflow, ensuring that it gives them a reason to keep using it. So the ROI, that’s always coming up in the discussion. And firms want to understand the value early on in the process, even if they understand that realizing that full value takes time, but it’s really in how they’re gonna implement it and how they’re gonna get the attorney to keep using it. Security is a major part of every discussion when speaking with enterprise firms. Again, it’s less of if they’re holding back because they’re not comfortable with generative AI.

I think it’s more that security and innovation teams, they’re being asked to evaluate more technology now than ever before. So those reviews take time. And I’ve worked with firms that genuinely are excited about a product, but it can still take months to get through that security, that procurement, and just other approvals that are required to actually deploy the tool. So for firms, it is less about trusting the AI and more about how to responsibly evaluate it and approve the tech that’s coming at such a fast pace for the firm.

Greg Lambert (23:01)
Yeah. Ross, I know you know, you were well before the AI transformation, but you were still a technology that was coming in. And I think, you know, watching your company over the years, just kind of the trust level that you’ve earned, was that a worry that through the AI era that might change for you and how did you fight that?

Ross Guberman (23:28)
Yeah, that’s a really great question. I mean, I’ve been lucky to enter the tech world with a reputation already for trustworthiness on the subject matter, right? That doesn’t necessarily mean it translates into trustworthiness on tech, but that did help. Because it also means that not only do I understand legal writing and I am reliable for legal writing advice, but I totally understand law firms and how they work and how they think. So that did help, especially when we were a fully deterministic product, as I mentioned, just code. When it comes to Gen AI, I think it helps on the margins, right? Because again, I do understand the security pressures that firms and individual attorneys face.

I mean I understand them at the core of my very being, not just intellectually. But, you know, as Heather suggested, we, you know, we face the usual gauntlet of questions and questionnaires. And although no legal tech company enjoys that process, we all believe in it and we all understand the concerns that these firms have. Luckily, I’ve noticed the sort of anxiety level has plummeted in the last couple of months. I think firms have kind of hit their stride. I mean they’re able to reassure their clients that they have the right type of account, that data is not being shared and the like. So it’s a little easier now than it was a couple of months ago.

But although again, I’m very glad that, you know, firms and judges and courts trust me as an individual, we still have to prove, as Heather suggested, we still have to prove our security bona fides just as any other legal tech company does.

Marlene Gebauer (24:52)
Okay. So I want to talk a little bit about the acquisition of WordRake. So WordRake was also a very established legal writing product. And I think BriefCatch and WordRake had kind of different approaches to improving lawyers’ work. I’m curious, how is the WordRake acquisition impacting the direction of the company and sort of the direction that you’re going?

Ross Guberman (25:37)
Yeah, I mean it was my first acquisition or our first acquisition.

Marlene Gebauer (25:41)
Congratulations.

Ross Guberman (25:42)
Thank you. Yeah, I didn’t know what to expect. It was kind of fun to be a client of law firms, since I used to be a lawyer at one of those firms. So the acquisition itself was, as I think these things go, like very, very smooth, very positive. You know, one of our star employees, in fact, right now at BriefCatch came from WordRake. So the acquisition and the integration in general were quite pleasant. I have good memories of them. It’s been almost a year. It was January, so you know, nine months or so. As far as the two startups, Gary Kinder is the founder of WordRake.

He and I generally think alike about legal writing itself. WordRake had a slightly different focus and a different kind of tech stack. It mainly focused on conciseness and plain language. So there’s some overlap, but then we have a lot of other layers as well. So we have a lot of former WordRake clients, of course, they are now BriefCatch users. They’re very, very happy because we have these AI features as well and other edits, but it was a really good synergy to use a trite word for mergers and acquisitions, but it’s true. And you know it’s been great because they also had a kind of a different customer base, you know, they were strong in different areas in the market.

So it’s been a really great transaction for us. And actually, we can’t reveal anything right now, but we’re looking at a couple of other possible acquisitions and hope that we’re on a roll since the first one was quite

Greg Lambert (27:13)
Mm-hmm.

Ross Guberman (27:14)
Quite successful. So stay tuned. I don’t think even Heather knows this. No one knows it. It’s all in my head.

Greg Lambert (27:16)
It gets addictive. Gets addictive.

Heather Rodriguez (27:18)
Yeah.

Ross Guberman (27:21)
But, yeah, we are looking at some other startups right now.

Greg Lambert (27:25)
Well, maybe this is leading into the question I’m about to ask then. Because you know it used to be BriefCatch was a Word-based, you know, plug-in that you used in Word and you know, the old workflow where the lawyers are working in Word. And now you’re looking at a much broader platform where you’re encompassing the writing, the verification, the training, the enterprise administration of it. So what changes when legal technology, writing technology stops being just another tool where the lawyers are working and actually becomes part of the infrastructure in which the legal work gets produced? How’s that change for the law firm and how’s that change for you?

Ross Guberman (28:09)
Yeah, I mean, great question. I mean, just an example. Like RealityCheck is really, really popular with the nation’s courts, but they don’t work in Word. I mean, they work in Word on their opinions, but that’s not how they process briefs. Right.

Greg Lambert (28:19)
Please don’t tell me they’re still in WordPerfect 4.2.

Ross Guberman (28:22)
No, no. There are a couple of judges who are WordPerfect users. They tend to be very proud.

Greg Lambert (28:25)
Yeah.

Ross Guberman (28:27)
But I think probably a handful, max, of WordPerfect stalwarts. So yeah, that’s an example, right? So when judges get a motion or brief in their docket and they want to run through RealityCheck, that’s obviously gonna be a PDF. And they don’t want the output in Word, right? They want it as a report. So to your point, it’s exciting and challenging, right, because one good thing about the original setup was it made things very easy for us, right? Just as our customers wanted to work in Word, all of our work was within the Word plugin. And now we have all these other layers and all these other plans and features on our roadmap, including web-based tools, Outlook, a web-based administrative platform, and web-based tools outside the platform for courts.

It goes on and on and on. But I will tell you, even with all that in mind, it is still true in September, October 2026, that the vast majority of lawyers are still doing most of their actual legal work in Word. Not WordPerfect and not Google Docs and not online. At the same time because we’re trying to capture a larger segment of the entire life cycle of a legal document, including the verification step and the review steps. We’re working on court formatting rules and the like. We are getting the opportunity to add to our tech stack while still letting our very content users do most of their editing in Word.

Greg Lambert (29:56)
I have to ask, the key thing is: Do you have an MCP yet?

Ross Guberman (30:01)
We actually haven’t integrated, you know, with Claude or OpenAI with an MCP yet. We use an MCP for RealityCheck in the other direction. But you’ve kind of probably read our minds because that’s you know, we have a long list of things we’re working on right now, but that’s you know, we’ve gotten some interest from those companies and that might be happening before the end of the year.

Marlene Gebauer (30:22)
So Ross and Heather, I want to pose the question to you. What’s true today that wasn’t true a year ago?

Ross Guberman (30:29)
Heather?

Heather Rodriguez (30:30)
I’d say that we’ve talked a bit about this, but AI is much more normalized within the legal industry than ever before. I remember being at a conference not even a year ago, and there was still a lot of hesitation. Firms had significant reservations, and I think now the conversation feels very different and there’s much more acceptance. It’s just gonna be part of how legal work gets done. For me, firms and attorneys have more of a hands-on experience now, so they’re much more comfortable and have a much better understanding of not only what it can do, but where it falls short. And so I think the experience has made the conversation much more mature, with less fear around technology and more recognition that it’s a tool.

And so it’s less hesitant to touch it and it’s really learning how to work alongside the attorney.

Greg Lambert (31:18)
Any observations of what’s true today that wasn’t true a year ago?

Ross Guberman (31:23)
I mean, Heather was about to say she now has the world’s greatest boss, which she didn’t have a year ago. So I took the words right out of her mouth. But yeah, that may not actually be true, so I won’t push my luck. So

Heather Rodriguez (31:28)
It is true.

Ross Guberman (31:33)
What I’d say if you’re a founder or you feel like you’re at the vanguard here with legal tech, one thing is there’s legitimate excitement almost every single week. There’s something in the legal tech or frankly broader Gen AI news to get you thinking, to get you creative, to get you thinking about possibilities that didn’t seem within the realm of possibility even let’s say a month ago. So that’s mostly really exciting. It keeps me going since I don’t sleep at all. It helps me

Heather Rodriguez (32:04)
Yeah.

Ross Guberman (32:05)
With my energy level 22 hours a day running this company. But it can also be, I think most legal tech founders would tell you, overwhelming, right? Even when you asked about MCP, I mean that’s one of like 50 things I think about every day that we need to get going. So I assume that this pace of innovation will ebb and flow, but right now it seems to be pretty maxed out, which as I said is mostly good but overwhelming I think for us and then probably overwhelming for the legal tech customers as well.

Greg Lambert (32:39)
All right, so that brings us from the past to the present. Now I need you both to pull out your crystal balls and look into the future for us. From your vantage point on the changes that you see in the way that lawyers you know create, evaluate, communicate legal work, what do you think that we need to be prepared for that will be a change or challenge in the near future? Heather, you want to start it off?

Heather Rodriguez (33:07)
Yeah, I’ll start off. I think we’re underestimating how intentional law firms are gonna become about building their own internal AI ecosystems. I think there’s a lot of focus on consolidation and finding the right technology to solve multiple problems. So I think there’s a lot of questions about if the future is necessarily one AI platform. I don’t agree. I think in three to five years it’s gonna be more of which AI are we using and how do we want AI to support the attorneys at each stage. So, yeah. I think that law firms are gonna be thinking about creating their own ecosystems and the right technology to do the right job and just making the attorney better at their own job.

Greg Lambert (33:50)
Mm. And Ross, what do you see in the future?

Ross Guberman (33:53)
I think an underreported story is how much the nation’s courts, you know, federal and state alike are quietly experimenting with all sorts of Gen AI tools within the courts. Like the only thing you hear about is sanctions for hallucinated cases, and it’s really misleading. And we’re very lucky because, again, especially with RealityCheck, but even with BriefCatch, we’re already, we’ve already penetrated the courts fairly well. And I think probably less than a year, but definitely in a year, what’s probably going to happen is the courts are going to be using Gen AI tools, 10 times more than they are today, and that is going to really open the floodgates on the adoption front.

I think the courts would be seen as sort of the last sort of source of resistance. Once they send the signal that they think overall this is good for our profession, good for access to justice, good for private practice and public sector lawyers alike, then it’s not going to be really a question of whether you use these tools. It’s going to be really which ones you use. And I think as exciting as legal tech seems right now, it’s probably going to be 10 times as exciting in a year, if my prediction comes true at least.

Greg Lambert (35:05)
Are you excited to be involved in the legal tech industry at this time?

Ross Guberman (35:11)
This is probably the most excited I’ve been since I started seven years ago. Yes. A little tired some days, but very excited. I think we’re lucky.

Greg Lambert (35:20)
Well, you said you didn’t sleep.

Ross Guberman (35:21)
I think we’re lucky. We’re lucky to be living through this, to be honest.

Greg Lambert (35:25)
Yeah, I feel the same way. So well, Ross Guberman and Heather Rodriguez from BriefCatch, we’ve dropped Marlene with technical issues. But we wanna thank you both for taking the time to come on and talk with us on the Geek in Review.

Ross Guberman (35:40)
Thanks so much, really enjoyed it.

Heather Rodriguez (35:40)
Thanks for having us. Thank you.

Ross Guberman (35:43)
Say bye to Marlene from us.

Greg Lambert (35:43)
And I’ll cover Marlene’s part. And thanks to everyone for listening. If you enjoyed the show, please share it with a colleague. We’d love to hear from you. Check us out on LinkedIn and Substack and other places. Ross and Heather, what’s the best place for listeners to find out more about you and BriefCatch and follow your work?

Ross Guberman (36:04)
I’ll keep it simple, BriefCatch.com. But what do you have, Heather?

Heather Rodriguez (36:08)
Follow us on LinkedIn, Ross Guberman and Heather Rodriguez.

Greg Lambert (36:12)
Right. And as always, the music you hear is from our friend Jerry David DeCicca. So thanks, and bye, everyone.

Ross Guberman (36:18)
Nice.

Heather Rodriguez (36:19)
Bye.

Ross Guberman (36:20)
Thank you.

For the better part of two years, law firms have been buying just about everything AI. Every vendor I talked to over that time expected renewals and expanded licenses at every firm. Every firm I spoke with said they were licensing widely and would cut back once they figured out what fit. Those two things can’t both be true. So I gave the vendors all the same warning: “You’ve only got a few years. At some point somebody in a partner meeting is going to ask what the firm got for the money, and if they don’t see clear ROI, you’re on the hook.”

I was right about the clock. I was wrong about who would end up on the hook.

Over the last few weeks, every one of my clients with a KM function has started some version of the same reorganization. The KM team, the Innovation team, or both are being folded into an IT-led AI group, often under a newly minted Chief AI Officer. My client list doesn’t make for an industry trend. But when I started calling friends at other firms, I heard very similar stories about firms I don’t work with.

Putting KM and Innovation under IT may not be optimal, but it isn’t necessarily bad. I started in IT, and my company works with several innovation teams that live in IT. This is not an anti-IT rant. But the trend does raise one big concern for me.

My concern is that firms are undervaluing knowledge, and in some cases sidelining the teams that manage it. We’ve spent two years with firms exploring which AI tools to buy. Now that many of them have made that choice, they need to shift their focus to telling those tools what the firm knows. Instead, they’re handing AI to the people in the firm who procure and manage technology. KM is how firms know what they know. The firms that win the AI era won’t be the ones with the best tools or the biggest budgets. They’ll be the ones with the best Knowledge Management feeding into their AI.

Why now

Four pressures are converging at this moment.

First, AI spending has gotten big enough that firm leadership wants one person to own it. One budget line, one org chart box, one throat to choke. And the logic for who that person should be is simple. AI is technology. IT manages technology. Therefore, AI belongs to IT. Both premises are true. The conclusion is still wrong.

Second, many contracts from the first wave of AI tools are coming up for renewal. That means the first real reviews and the first real ROI analysis. For a lot of AI purchases, clear ROI is hard to point to. Adoption is uneven. Some time savings are real but hard to measure, and firms are learning that AI use doesn’t automatically equal time saved. And in a business that bills by the hour, saved time isn’t obviously a win.

Third, it’s budget and review season. If a firm is going to redraw its org chart for next year, now is when it happens. Budgets get set, headcount gets allocated, and someone gets named the owner of next year’s AI line item.

And fourth, very few law firms have original ideas. When one firm names a CAIO and restructures KM and Innovation under them, every managing partner starts asking why their firm hasn’t done the same. Nobody wants to explain to the partnership why they’re the last firm follow suit.

Each of those reasons makes complete sense on its own. Together, they look like an AI strategy. They’re not.

Where I got it wrong

I assumed a large number of vendors would take the hit when their products came up for renewal, and they may still. But vendors aren’t in the partner meeting when someone asks what the firm got for the money. The KM or Innovation leader is. They championed the tool. They picked it, ran the pilot, promised adoption, and wrote the business case. When it doesn’t deliver, for any reason, they own that.

The vendors got a reprieve. Their internal champions are paying for it.

KM isn’t blameless here. Too many KM and Innovation teams have focused on procurement, chasing vendor demos and conference swag instead of their lawyers’ needs. And frankly, KM has never been great at explaining its value to the people who hold the budget. That said, I don’t see any connection between how well a team performs and whether it’s being reorganized. And failing to make the case is no reason to take KM out of the equation.

The wrong moment to make that call

For the last few years, treating AI like any other technology procurement wasn’t entirely crazy. AI is definitely technology. We’re not exactly sure how it works. It keeps changing. We don’t know how we’re going to use it. Let’s just get it in here and figure it out as we go. If AI means licensing a legal AI platform, rolling it out, and training people to use it, then it looks a lot like buying a document management system. Pick the vendor, negotiate the contract, pass the security review, deploy, and measure adoption. That’s the standard playbook that KM and Innovation teams have been running for two years. Unfortunately for them, every step of it is squarely in IT’s wheelhouse. For management, that makes the reorg decision an easy call.

But that’s not where the technology is heading. And I don’t think anyone, including me, has done a good job of explaining to firm management what comes next. Hence their Underpants Gnome approach to AI:

Phase 1: Buy AI.

Phase 2: ???

Phase 3: Profit.

The tools that will matter going forward are agents, not chatbots. They sit on a lawyer’s desktop, read the files, and carry out multi-step work on their own. What will take those tools to the next level isn’t the model, the interface, or the infrastructure. It’s context. Does the agent know which precedent the firm relies on? Which positions the practice group takes, and which it abandoned years ago? How a particular partner wants a particular document built? What this client cares about? None of that is in the model, and none of it is in the vendor’s product. It lives in the firm, scattered across people, documents, and habits. KM is the only group whose job has ever been to capture it, curate it, and make it usable.

The people in KM think of themselves as knowledge professionals, not technologists. At a law firm, many of them are lawyers or former lawyers who never had any interest in working in IT. To them, the move looks like a step backward in their careers, and they’re likely to leave. When they go, a lot of what the firm knows goes with them.

Your knowledge is your advantage

Large firms are all buying the same AI tools. And despite what the big vendors tell you, those tools are interchangeable. This one has a feature that one lacks. That one does a certain kind of work slightly better. Underneath, they’re the same. You need them to stay in the game, but buying one of them, or all of them, gives you absolutely no competitive advantage.

Your firm’s knowledge is what can make AI a competitive advantage. And giving your AI tools an MCP connection to your DMS and other systems of record doesn’t give them your knowledge.

Your systems of record store documents and data. Your firm’s knowledge is bigger than that. It’s knowing who has handled this exact issue before. It’s knowing how a matter actually gets staffed and run. It’s knowing which deal went sideways and why. It’s knowing which argument the judge didn’t buy.

Much of it was never written down. So an agent with access to every system you own still won’t find it. Somebody has to capture it. Somebody has to keep it current. Somebody has to put it where an agent can use it.

That somebody is KM.

KM is Phase 2.

And without KM, underpants never turn into profit.

If your firm has recently reorganized in this way, or is in the process of reorganizing like this, your KM people are probably still in the building, for now. Go find them and beg them to help you build a Context Strategy.  The firms that don’t will figure it out eventually and they’ll have to rebuild KM from scratch. But if you start now, you should have a two-year head start over your competition.

Law firms and legal departments face pressure to adopt AI, yet buying another tool does not answer the hardest question: Which problem needs solving? Dr. Maryam Salehijam, founder and CEO of Andarzi, joins TGIR hosts, Greg Lambert and Marlene Gebauer, to discuss how legal teams choose between existing systems, new vendors, and tools built for their own workflows. Her starting point is to observe the work, identify a useful outcome, and then decide what technology belongs in the process.

Salehijam describes legal teams paying for AI platforms while struggling to get consistent use from them. She urges organizations to examine tools they already own and ask vendors for more training before adding another subscription. A legal department’s use of ServiceNow for intake provides one example. The conversation also tackles shifting model capabilities, consumption pricing, and the cost of assigning an expensive model to a routine task.

What counts as a return on legal AI investment? Time and money matter, but Salehijam argues for measuring employee satisfaction as well. Repetitive contract review and document work consume attention lawyers would prefer to spend on judgment, business problems, and client relationships. She describes using anonymous feedback before and after a workflow change to learn whether people’s work has improved, while Greg raises the tension between hours saved and the billable hour.

The discussion turns to local AI models, data security, and the growing interest in AI-native law firms. Salehijam argues for decisions based on the task and the client’s requirements, including a direct conversation with clients about how their information should be handled. Greg presses her on whether smaller firms and large firms face different constraints. Marlene asks whether the emerging “legal engineer” role will endure, prompting a pointed exchange about technical skill, credentials, and what lawyers need to learn.

For teams putting AI into practice, Salehijam favors hands-on learning with peers, repeated use, and work product over a quick certificate. She also warns leaders to account for the time and effort a new workflow demands before promising immediate gains. In the closing questions, she reflects on a legal AI market with fewer new entrants than she expected and predicts more law firms and legal departments will ask whether owning a workflow serves them better than renewing another set of software seats.

Listen on mobile platforms:  ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Apple Podcasts⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ |  ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Spotify⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ | ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠YouTube⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ | ⁠Substack

Email: geekinreviewpodcast@gmail.com

Music: Jerry David DeCicca

LINKS

• Andarzi
• Dr. Maryam Salehijam on LinkedIn
• AI Fluency for Legal Professionals on Maven
• Harvey
• Legora
• ServiceNow Legal Service Delivery
• Notion
• Astra for Law

Transcript

Marlene Gebauer (00:07)
Welcome to the Geek in Review, the podcast focused on innovative and creative ideas in the legal industry. I’m Marlene Gebauer.

Greg Lambert (00:14)
And I’m Greg Lambert.

Marlene Gebauer (00:16)
This week we’re joined by Maryam Salehijam, founder and CEO of Andarzi. Maryam started on the technology side of legal way back when AI mostly meant e-discovery and moved to the services side when the technology wasn’t ready and came back when generative AI arrived. Along the way, she held roles at LexisNexis, Axiom, Robin AI, and Junior. Through Andarzi, she helps law firms and legal departments figure out what problem they actually have and what to do about it.

Greg Lambert (00:46)
Yeah, I like the good old days of when AI meant e-discovery. So, Maryam, welcome

Marlene Gebauer (00:51)
Ha ha ha.

Greg Lambert (00:52)
to the Geek in Review. Good to have you here.

Dr. Maryam Salehijam (00:53)
Thank you, Greg. Thank you, Marlene. It’s my pleasure to be here today to speak about my new initiative, Andarzi.

Marlene Gebauer (00:59)
So Maryam, we mentioned the places you’ve been in where most of us know you from, but you recently launched, as you said, your new company, Andarzi. So what inspired you to go out and build it?

Dr. Maryam Salehijam (01:09)
That’s a great question. Well, it honestly came from encouragement from my family and friends. They all said we don’t know why you don’t start your own company. If there’s one person to have her own company it’d be you. That’s literally what it was.

Greg Lambert (01:19)
Stop making other people money and make yourself money, right?

Marlene Gebauer (01:23)
And I’m sorry, like if your family is already like involved in AI and they’re talking about that too.

Dr. Maryam Salehijam (01:30)
Yeah, they said go for it. but really the reason I jumped in is because I kept getting calls from my friends in legal departments and law firms saying, “Hey, you’re in AI, right? And I said, “Yes.” And they said, “We don’t need what you’re selling because we do this other thing, but can you help us with our strategy? We’re trying to figure this out. There’s all of this pressure. If you’re at a legal department from your management at the company, the C-suite. If you’re at a law firm, it’s from your clients. We don’t know what to do with it and we don’t have the time to figure it out on our own. We need somebody to come in who we can trust to tell us what’s going on, where does AI make sense, where does AI not make sense and what should be our strategic plan for the next six months or year, right? What’s gonna happen to us as an industry and that’s where Andarzi was born from. And the name actually comes from the Persian word “andarz”, which means to give advice.

Greg Lambert (02:21)
Very, very nice. And so when they reach out to you with initially, what’s some of the common problems that they’re having that is causing them to reach out to you?

Dr. Maryam Salehijam (02:33)
That’s a great question. So let’s talk about, for example, a decent sized legal department where they don’t have a huge internal engineering team inside of the legal department. Their main issue is they’re getting this mandate to become more efficient, spend less on law firms, right? Hire less. But they don’t really know what tools to choose from. Let’s say they wanna deploy a tool for contract intelligence. There’s at least two hundred tools out there. They all say the same thing.

Greg Lambert (02:59)
Sure.

Marlene Gebauer (03:00)
Ha ha ha.

Dr. Maryam Salehijam (03:01)
They all tell you they’re the best and the pricing varies. How do you choose the right one for you? Right? The second issue is if you’re buying multiple tools, how do you keep on top of everything? How do you keep on top of the training? How do you get your people to use it? I have so many friends inside legal departments at companies that bought from the likes of Harvey and Legora, and they say, “Maryam, our utilization is not where we need ito be for this to make sense. Right? We’re really worried. We’re paying a lot of money, but not really seeing the savings or the efficiency we would hope to see when we went into this. And then, for the law firms who don’t have an internal chief innovation officer or really a team dedicated to figuring it out, they’re going, Well, if I try to buy a tool for every problem I have, I’m spending north of $300,000 a year. And that just doesn’t make sense for me so could I potentially build some of this myself? And I see a lot of conversation coming in of is this something we can build? And so that’s where I go in and say, okay, wait a minute. Before we make a decision, let’s actually figure out, let’s observe your workflows, where is there a quick win for you that you can have with your existing systems? Sometimes they really are underutilizing what they already have. And then where it makes sense we can make the determination between build versus buy. And guess what? If you want to buy I can help you With all the vendor selection, I can help you negotiate those contracts. And if you want to build, I do partner with two agencies that can come in and do the AI build for you.

Greg Lambert (04:23)
Very interesting.

Marlene Gebauer (04:25)
Mm-hmm.

Greg Lambert (04:26)
one of the things that when you and I were talking in prep for this, we were talking about how some of the firms are still focusing in on the prompt engineering that it’s still around the prompt around the chatbot. So what are you seeing, and what are you advising firms and other legal organizations to build and training on that’s beyond the prompt itself. What are you telling them to start focusing in on?

Dr. Maryam Salehijam (05:01)
What I’m telling them is to throw everything they know about technology out the door because the speed at which Gen AI is moving is beyond comprehension. So maybe four months ago we could have said learn how to prompt that’s important, that is no longer a valid skill to have because the systems can prompt themselves. So the way I think about it is think about where cars started where you had to hand crank to get it going and it was really unsafe and there were so many things that could go wrong. To now we have self-driving cars in the shape of Waymos, not even a human in the seat. I mean, they’re also really sketchy if you ask me, but that transition took what 200 years? Gen AI is doing that kind of transformation within less than 12 months. So when it comes to skill set for law firms and legal departments, I really encourage all of my clients to think like they’ve never done before. It’s not like law school where you go to law school and you’ve learned. It’s very much every day there’s a new thing that’s coming out that’s gonna scrap everything you’ve learned about how you know how to do things. And you have to have that mentality to really stay on top of everything.

Marlene Gebauer (06:08)
Yeah, it’s really challenging because, you know, you’re talking about we’re talking about prompt engineering, right? And at one point, like maybe last year, that was actually a job. And now that is not a job anymore. And so for firms trying to plan for, you know, hiring, you know, allied professionals or even trying to determine what kind of skills you know, attorneys should have in their organization. It’s it’s really rather challenging. what kind of feedback are you getting?

Dr. Maryam Salehijam (06:39)
exactly that they feel like they’re constantly behind. Everyone feels behind at all times, which is what I tell them is like actually you’re in the same boat with everyone else, so you’re not behind on anything everyone feels behind. So don’t stress it too much, but also don’t jump. I always tell people don’t jump into hiring people if you’ve never had someone do this for you. Start with a consultant, start with a contractor. If you are not a massive organization, if you are not a massive law firm, if you are not a massive legal department, you do not need a full Time person doing this kind of work for you. And guess what? A lot of your vendors will do most of the heavy lifting. So you just have to ask them. So many people don’t actually talk to their vendors. They buy the product and then they go and they start trying to figure out how to use it internally. And I go, no, the vendor wants to work with you. You’re just afraid of working with them because you think they’re gonna sell you more stuff. Just learn to say no and use your vendors for what they’re good at, which is teaching you how to use their tooling. so that’s kind of the advice I go in with. Again, it’s super practical, right? I’m like, don’t spend any money. try to use what you have try to go to the people learn to say no right yeah don’t spend

Greg Lambert (07:39)
You’re speaking the right language.

Marlene Gebauer (07:41)
Okay.

Dr. Maryam Salehijam (07:42)
anything and that’s like one of the legal departments I love the most because they used to be my client over at Axiom Nutanix that’s their mentality as well they said before we buy a legal intake tool for the front door of our legal department well we have ServiceNow can we use ServiceNow for this and they engineered ServiceNow to do exactly what they needed to do and they did not have to spend hundreds of thousands of dollars to buy a tool to do it. Right? And it’s that kind of mindset that I think is gonna make legal leave the stone ages where we were always seen as not technical people to come into, hey, we’re actually at the frontier of this general AI movement because it makes sense for our business model.

Greg Lambert (08:24)
that’s let me let me kind of read back what I think I heard here. because I think a lot of us are thinking, you know, AI yeah, it’s been around three and you know, three and a half years, going on fourth year now. but it’s still pretty new and we think that we need to jump into whatever the newest and latest and you know the startups are. But from what I’m hearing, you’re saying we’re probably not leveraging our existing relationships enough. Is that is that what I’m hearing?

Dr. Maryam Salehijam (08:54)
One hundred percent. I’ll give you another example. I work with a big pharma company where they simply need to cross-check three data points across fifteen different types of document. just project at a time. They tried to use their own internal tooling and they found it wasn’t working well. But it’s because they were simply prompting it. They didn’t know how to set up some sort of vault within their systems, how to build a skill set, how to build the citation checker, how to do the charting. But their existing Copilot tooling could do all of that. They just didn’t know it could, right? So it’s an education piece that all I had to bridge is the education. Hey, this is how you can do it. They did it. Voila. Didn’t have to buy a tool with a vault function at thousands of dollars just to do something Gen AI on its own if you already have access to Copilot or Google Gemini can do for you.

Marlene Gebauer (09:47)
Yeah, it’s amazing that the you like sometimes we have so many products that we don’t know what we actually have and what it can actually do. interesting. So all right, I’m I’m very interested in this question here because all we’re hearing about is consumption-based pricing. Consumption-based pricing. Consumption-based pricing. now you’re saying tokens are becoming a commodity. So Okay, that goes a little bit against the grain of what we’re hearing and what not so much what we’re hearing, but what seems to be a big concern among law firms and like how are we gonna pay for this. So what happens, you know, to the vendors and their pricing when that all plays out?

Dr. Maryam Salehijam (10:31)
That is a great question. And I actually had a like a mini debate with someone at the bar at running legal like the business because they were

Marlene Gebauer (10:36)
I bet you did.

Greg Lambert (10:37)
Yeah.

Dr. Maryam Salehijam (10:39)
at one of the vendors that does token based pricing and they were talking about how great that is. And I said, That is a terrible business model for your clients and he said, Why? I’m like, There’s no predictability on their bills. This is one of the reasons people get mad at law firms, right? You just send them a bill and they go, oh my God. What is c

Marlene Gebauer (10:55)
Right, like w one time it’s super expensive, then the next time it’s gonna be not as expensive.

Dr. Maryam Salehijam (11:00)
Right.

Marlene Gebauer (11:01)
And we’ve done this before, you know, we’ve done this consumption based pricing before and nobody

Dr. Maryam Salehijam (11:06)
Yes.

Marlene Gebauer (11:06)
liked it.

Dr. Maryam Salehijam (11:07)
Nobody liked it. And I also really dislike this whole thing in legal that I see of like, oh my God, Claude for Legal is the winner. now it’s OpenAI. Let’s write articles about it. Let’s

Greg Lambert (11:15)
Ha ha ha.

Dr. Maryam Salehijam (11:16)
move our skill set. Unlike everybody calm down.

Marlene Gebauer (11:19)
Mm-hmm.

Dr. Maryam Salehijam (11:20)
There’s no winners or losers yet. Let’s just not constantly make deterministic decisions. Again, that you d this is like never before. You cannot apply your same mentality. It’s either Lexis or TR. It’s either this or it’s that. It’s not like that. It’s not black and white. It’s very much gray. Even for the providers, it’s gray. They don’t even know what’s happening after time, right? So when it comes to tokens, we know they’re very resource intensive, right? From data centers, from the cost of compute, from the cloud. It’s very, very expensive. And certain tasks consume a lot more tokens than others. Certain models consume a lot more tokens than others. I also see this fallacy constantly where people are like, why doesn’t Lexis have the latest version of Claude for this? I’m like, because it doesn’t need to. You don’t need the latest models for everything, right? It’s that education piece where we need to really think about what is the task, how simple is this task, right? Or how complex is this task, and what is the cost of utilizing AI on this task for each project? Before we try to,

Marlene Gebauer (12:21)
And is it worth it?

Dr. Maryam Salehijam (12:23)
is it worth it before we try to throw AI on everything? And then my friends who actually work at the Frontier Labs and the actual tech companies, so many of them are moving to local compute power where there’s basically no cost to the token. And that’s the other conversation we need to start having of okay, why not just do it locally? It’s actually more secure a lot of the times. It’s cheaper. and from a technical perspective, also not that hard. So how do we also in think about that as law firms and legal teams if our one of our biggest concerns is secure? Right? Before we just go spend more money buying stuff. So it’s a lot of those conversations of let’s first actually understand what we have available to us before we make a decision on what we’re gonna do with it.

Marlene Gebauer (13:01)
Have you heard any discussion about, you know, some of the some of the vendors are using you know their training on models that are like foreign models, like Chinese models. Like have you heard?

Greg Lambert (13:14)
Yeah.

Marlene Gebauer (13:16)
oh my God, we have to get the eye roll. That was beautiful. You missed the eye roll.

Greg Lambert (13:18)
I was gonna say for the for the for the people that are only listening, you missed probably the best part of the show.

Dr. Maryam Salehijam (13:25)
oh my God.

Marlene Gebauer (13:26)
So wha you know, what are your thoughts on that or what are or what are you hearing from your clients about vendors that are doing that?

Dr. Maryam Salehijam (13:34)
Yeah, great question. So I think everybody wants to sound like they’re going against the grain because it’s cool, right? oh my God, we’re doing it this differently. It’s cheaper, it’s

Marlene Gebauer (13:43)
It’s like it’s cheaper. Okay, it is cheaper.

Dr. Maryam Salehijam (13:45)
this, that. But actually, like this is the other thing where again some of the vendors are gonna hate me after this because I just say what I really think. Why are you trying to build your own LLM? You think you have more resources and knowledge and money than OpenAI or Anthropic? Are you kidding me? It’s never gonna be better. Like, it’s just those things where I’m like, so you’re gonna go trust a Model that hasn’t been vetted at all for our industry for these things. You’re the person vetting it, which doesn’t matter. I don’t care about your personal benchmarks. That’s like me being like, I’m the greatest. Doesn’t matter. my mom thinks I’m great. Doesn’t matter. Against literally companies that have so much money and resources behind them that they can surpass the GDP of a country. Like, let’s get real here. This is this is not logical, right? Why would you do that? So it’s a lot of those things where I go, oh my God, here we go again. Somebody’s like saying something that makes no sense. Like nobody has more resources than OpenAI, Anthropic, or Google. Maybe Microsoft, right? Maybe IBM, but right now, if you’re not plugging in OpenAI or Anthropic, I don’t really think you’re that serious about understanding Gen AI.

Greg Lambert (14:55)
I wanna dig a little deeper on something that you mentioned earlier and that was using local models. are when can you define that for me? Is that something that’s running like literally on your laptop or is it running in your cloud? Well can you define

Dr. Maryam Salehijam (15:13)
Yeah.

Greg Lambert (15:13)
that a little bit more?

Dr. Maryam Salehijam (15:15)
Is running on your machine, or even your phone has compute power, or you can plug in a GPU from NVIDIA or Apple. And So what happens when we need to process data, right? We need compute power. Now you can have compute power via the cloud, but actually the cloud ultimately connects to the data center, or you can have a mini version of your own data center connected to your device. Now, a lot of smaller law firms are already doing that actually. the one-person shops, right? Especially in The IP space because it’s super secure. Data’s going nowhere. Nobody’s seeing the data. But that only works if you actually know what models to select, what the task is, you know how to quality check. It’s not gonna be something you can easily use for general purpose things, like it’s not gonna replace your OpenAI or you know Anthropic subscription. But let’s say you’re constantly reviewing a specific type of contract where you know exactly what to do, you’re writing a specific type of document, it’s perfect for that kind of work because you know what the task. is you know what the model weight needs to be, how complex it needs to be, and you know what the deliverable looks like. So then it’s quite actually easy to do it yourself. The same is true for anything that contains like patient information, right? Medical records, timelines, easy to do it locally, safer. You don’t have to worry about HIPAA compliance.

Greg Lambert (16:30)
And are you seeing like smaller companies, departments that are experimenting with the with the local GPUs to run this?

Dr. Maryam Salehijam (16:40)
Yeah, not and not just the small, right? Latham, I think just announced they bought a whole GPU farm to bring in house. So it’s happening at all levels and I will suspect that’s gonna happen more and more as we struggle with this concept of when do we wanna use AI and not only when, but how do we wanna use that AI. Are we getting our water in a bottle? Are we getting it through the tap? Are we getting it delivered via a truck? How are we gonna use it? What makes sense? for us and what makes sense for the specific task. So I really think people are gonna start thinking more efficiently also about where they’re gonna get their AI from and why.

Greg Lambert (17:20)
So I think if I look back at my predictions for 2026, I think one of the predictions was that law firms and legal departments were going to have to come up with the return on investment and the metrics of you know, we’re we’ve spent so much money in experimenting with AI, but at some point we’re going to be asked about how do we measure the success on that. So what we’d love to measure, you know, it’s like hours saved, money saved, I guess, but what do you think and what are you advising your firms and legal departments on what they should be measuring to show the return on the investment that they’re making?

Dr. Maryam Salehijam (18:06)
I love that question. First of all, let’s talk about the elephant in the room. Law firms measuring time saved.

Greg Lambert (18:13)
What could go wrong with that?

Dr. Maryam Salehijam (18:13)
What?

Marlene Gebauer (18:14)
Yeah.

Dr. Maryam Salehijam (18:15)
Absolutely like logical, everybody wants to do that at a law firm. Don’t worry about billables. Like again, it’s just those things where I’m like, what are we talking about here? but I actually have something that I say which is so against the grain. I talk about happiness. One of the biggest problems in our profession is not how many hours it does take, is that depression, anxiety, burnout. It’s insane and legal. Why are we so miserable? I’m pretty sure none of us went to law school to be miserable. and we have one of the highest rate of suicide.

Greg Lambert (18:43)
Mar Marlene did. She went to law school in Cleveland. So

Marlene Gebauer (18:48)
E everybody in Cleveland I want you to know that Greg said it, not me.

Dr. Maryam Salehijam (18:52)
That’s so funny. No, Cleveland has a good law school. Come on now. I’m sure it was fun. Yeah.

Marlene Gebauer (18:57)
At Cleveland? Yeah, it does.

Dr. Maryam Salehijam (18:59)
but that’s like can we focus on what we really wanna do in legal, not the work we don’t wanna do? Who wants to review an NDA? Come on now. Who wants to do this? Who wants to redline the same provision in the contracts over and over again? Nobody wants to do that. Nobody wants to do doc review. At some point you just lose where you’re at. So it’s just all of this work that I actually consider a poor use of the human brain. It’s not critical tasks. Let’s have these systems do these things for us so we can focus on what really brings us fulfillment. And by being fulfilled, we actually start to get more creative. And we start to solve the business problems and we starto be better partners, not just to our clients as law firms, but also to the clients of the legal department at a company, right? Because at the end of the day, it’s the business that’s your client. And we can become more business people than legal people, because so much of law right now doesn’t consider business, technology. And I think that transformation is gonna be good for our industry and yes it might threaten the billable hour but that doesn’t mean it’s gonna wipe out law firms they just have to rethink how they generate their money, right? And potentially the split of their revenue and where it’s gonna come from and maybe law firms start offering more of the managed service like Eversheds does or Flex by Fenwick does, right? Where they have the actually multiple streams that they can offer their clients outside of just billables for specific projects.

Greg Lambert (20:25)
Yeah. How I know this is a tough question to ask you. but you know it’s easy to measure dollars saved, it’s easy to measure time saved. when you talk about job satisfaction, if you talk you know, those sorts of things, how do you measure that and are there other industries or examples that we could look toward that would help us kind of wrap our minds around how you measure that and what it looks like.

Dr. Maryam Salehijam (20:56)
Absolutely. So you can do the most basic old school way, which is employee satisfaction surveys, right? They’re anonymous. You can survey them before, then a year later. Or actually the company I partner with, what we do before we deploy on a project is we send out voice agents that are conversational, you talk to them. But it’s super anonymous. And people just talk to them about their workflows, what they’re frustrated with, what they like, what they wish they could do, what they wish they couldn’t do. And then we aggregate all that data and build a couple of solutions for them that we think would address the main things people are annoyed about. And then we deploy and then do the voice agents again to see are the problems solved or people happier, what’s going on. and data, you know, even though happiness is hard to measure, data can help you measure that, right? How do people feel? Do they feel fulfilled? What’s going on? I can tell you for in-house teams, the biggest pain point is often their workload. It’s way more work than they anticipated. They don’t have the headcount, right? And so sometimes they end up leaving their company and then that’s so expensive for the company to try to replace them. It’s hundreds of thousands of dollars, right? That retention issue is huge for companies. And with all the layoffs, oh my God, additional mental load. So it’s about creating that safe space for people to do their work, but do their work enthusiastically and as efficiently as they can.

Marlene Gebauer (22:12)
I just think it requires sort of extraordinary leadership and courage and creativity to sort of implement that kind of change management on a large scale because it’s so much easier to just say, Okay, we’re just gonna raise our rates and we’ll just give you a discount and you know, that’s kind of what we’ve been seeing this model for years and years and years, rather than actually fix the model.

Dr. Maryam Salehijam (22:38)
Yeah.

Marlene Gebauer (22:38)
And you know, I know that we have always said, well, maybe this is the time and maybe AI is gonna be what does it but I mean what are you hearing in terms of you know from about this like from clients and like how they’re how they’re trying to trying to do this?

Dr. Maryam Salehijam (22:58)
Yeah, that’s a great question. I think for a long time, businesses, right, they didn’t really know how to get legal work done without lawyers. Right? We were the gatekeepers, we had the knowledge, we had the know how. But now these AI models can go and give advice and I know people lawyers are gonna get mad at me, but a lot of the advice is pretty good. So a lot of the stuff it says and does is hmm, it’s pretty good. And so that grace that law firms used to have by their clients of well you’re the expert, you’re the professional, that’s going out the door really fast. And I think the reason the change might come now is because it’s no longer in our control how people get their legal work done. Companies are starting to do a lot of what they used to send to law firms in house. And so and I know law firms are still growing ’cause they’re increasing their rates, but is there actual work growing? I don’t know. and that’s where I feel like there might be a reckoning. Now, not for the work streams that you have to have a law firm, not for compliance, not for litigation, right? Certain work streams are safe. But if you’re a transactional attorney and the bulk of your client work comes from product counseling, privacy, commercial, labor and employment, just doc review, and not actual in court or anywhere else, you’re at risk.

Marlene Gebauer (24:15)
I wanna sort of go back to what you were saying before about kind of throwing AI at everything and that maybe we don’t need to do that. you know, maybe the fix is, you know, sometimes it’s AI, you know, sometimes it’s other software, a flexible legal team, kind of new staffing, or you know, no new technology. So how do you decide, which one a client needs or a multi you know, multiple of what they need based, you know, based on their workflow.

Dr. Maryam Salehijam (24:47)
Yeah. That’s a great question. So I’ll give you an example of a client problem I’m dealing with. They have no matter management system. It’s an Excel sheet. And they don’t know who’s doing what work. I mean, that one you might be like, let’s throw AI at this. But before we do that, we have to actually see what are they tracking in their Excel? What do they want to track? Where does the data live, right? Does any of the AI tools they want to use have those connectors? What is their budget? How much time do they have for implementation and training? How easy does the workflow need to be? And then based on all of that, then we can say, actually, all you need is a project management tool, and you can go buy a tool like Notion. And Notion has its own built-in AI workflows. Easy peasy, everybody uses for project management. Or you might need something like Clio because you need to do more complex stuff, but that’s pretty expensive. Or maybe we can vibe code this for you. It’ll be yours. AI is sometimes inaccurate, so you need to be careful around that, right? So it’s very much about showing them the end results and how many steps it takes to get there and letting them decide for themselves what do they want to do. Because when you make the choice of what to do versus what I tell you to do, you’re gonna like that choice a lot more. Now I can guide you with it. but the other thing I always tell people is if you’re gonna build it, you’re gonna like it more. And they go, What? I’m like, yeah, because it’s your baby. And nobody calls their own baby ugly. So you’re probably gonna use it more. So

Greg Lambert (26:17)
Yeah, yeah.

Marlene Gebauer (26:17)
Ha ha ha.

Dr. Maryam Salehijam (26:18)
it’s like I don’t wanna tell you to build. But if you build it, you’re probably gonna think it’s great. So that’s the other kind of like it’s just psychology at the end of the day.

Greg Lambert (26:27)
Yeah. Maryam, I want to kind of take what you were just talking about with things like you know, use Notion in order to do your project management and all. Are you dealing with any of the firms that consider themselves to be what’s called an AI native firm?

Dr. Maryam Salehijam (26:48)
Yeah.

Greg Lambert (26:48)
Cause when you when you talk about using products off the shelf like a Notion or a Cursor or other

Dr. Maryam Salehijam (26:54)
Yeah.

Greg Lambert (26:54)
tools, if I were to try and use that at a large firm, I would get pushback from my security ops people saying, you know, those are not secure, you know, the information that you’re using is not ours, it’s our clients’ information, and we have to be very careful with that. So One, are you finding that the firms that are calling themselves AI native or AI native adjacent, I guess, are able to use these types of off-the-shelf tools effectively, securely? And do you think that maybe the larger firms are overly sensitive when it comes to using some of these tools?

Dr. Maryam Salehijam (27:38)
Yes and yes. So overly sensitive

Greg Lambert (27:39)
Yeah.

Dr. Maryam Salehijam (27:40)
for sure. Some of these law firms, I’m like, what do you think you’re dealing with? National security? Like, this is crazy.

Marlene Gebauer (27:46)
Yeah.

Dr. Maryam Salehijam (27:47)
Like this is like the it’s like but also my question to law firms is have you asked your clients what they would prefer? Because guess what? Your client is throwing this the same stuff you’re holding near and dear everywhere. It’s on seven different Slack channels. It was copied and pasted somewhere. It’s on in some random drive. Like before we make such black and white decisions, let’s actually understand which clients need that crazy level of security. So for them we don’t put the stuff in the efficiency tools. And which clients would love for us to l use the efficiency tools. And guess what? If it’s a general counsel at a decent sized company that’s tech forward, they might want to get added to that Notion tab or any project management tool you use so they don’t have to email the partner and ask for an update and they can actually see how things are progressing, right? So there’s so much value we can add to our clients actually by having more visibility utilizing more tech, but we hide behind a security layer as if security’s so hard to figure out. At the end of the day, like outside of a patent litigation or really it like, you know, super confidential stuff, everything else we’re doing was emailed somewhere else at our client. So I don’t really know what we’re worried about. And like you can make things secure, but you don’t have to pretend like it’s the nuclear codes. It’s it’s a contract. Like no one’s like scraping the web trying to find that one contract. So again, it’s just sometimes I think we’re like scaring ourselves into staying in the stone ages for no reason.

Marlene Gebauer (29:18)
I wanna go back again. So we were talking about prompt engineers, right? And how that’s kind of gone by the wayside. Now sort of the title du jour is sort of legal engineer. And I

Dr. Maryam Salehijam (29:30)
Yeah.

Marlene Gebauer (29:31)
I think it’s a little different in that with you know, with a legal engineer, this is somebody who, you know, does understand the technology and can use the technology and you know, build agents and workflows and things like that. But They also understand the law. you know, they’re generally they’re lawyers that are doing this. And but I’m wondering, I mean, is this something that maybe is also a temporary type of job? Or is this something that’s always going to be, you know, part of the fabric and in a support role, or is this something that, you know, a legal engineer is essentially going to become what, you know, like But a lawyer is, like that’s that’s what everybody does. you know, what should what

Dr. Maryam Salehijam (30:13)
Yeah.

Marlene Gebauer (30:14)
should we should we be

Dr. Maryam Salehijam (30:16)
Yeah.

Marlene Gebauer (30:16)
looking for? ‘Cause I imagine they have to know law, they have to know data and what’s

Dr. Maryam Salehijam (30:21)
Yeah.

Marlene Gebauer (30:21)
important and technology as well.

Dr. Maryam Salehijam (30:24)
Okay, I love that question. So my gut reaction to the term legal engineer is ugh. ‘Cause you’re not an engineer. Like, why are you calling yourself that? What is going on? I did not know false

Marlene Gebauer (30:34)
damn.

Dr. Maryam Salehijam (30:36)
marketing was like on the table. “I’m a legal engineer.”. I’m like, and I took the course by Harvey, it was a joke. I’m like, this is the certificate. Like, this is not testing anything. This is what? and then people are like, I took it justo be like, what is this? It’s so easy. If anything, you should go through an official AI lab to get certified. But again, it’s wrong to call ourselves engineers when we are not engineers. We are not engineers. Engineers, we are not. I’m not an engineer. I’m a lawyer who understands technology, right? So you can call yourself a tech consultant. You can call it what it is. You are not an engineer, right? So that

Greg Lambert (31:13)
Yeah, but I can add another hundred thousand dollars to my paycheck if I call myself an engineer.

Dr. Maryam Salehijam (31:18)
an engineer. It’s very false marketing. so again I really care about like keep it to your core but also this concept of well it’s a lawyer so they understand a lot. Do they? I’m sorry most lawyers are specialized in their practice. Try to get an employment lawyer to talk to an M&A lawyer about private equity deals. They don’t know what the other person’s talking about. The terminology is different everything is different. The mindset is different. So that’s also weird to say. I think what you need right now in legal, and this might be just because I’m biased towards my own company is people who can ask good questions, understand your problems, who then can go ask good questions to other people, understand their solutions, and then come to you and say these are the ways you might be able to solve your problems. Decide what problems you want to solve. By the way, any problem you choose to solve is gonna suck at first because you have to spend time fixing it. Right? Anything that’s broken doesn’t get fixed easily. So that’s what we really need, but I think the title is just getting marketed around to make it sound fancy for some reason. What we should focus on as law firms and legal departments is ensure that every single person on our team is actually certified to use AI tools. And again, an actual certification that is hands-on and results in a work product where they’ve actively built something. Not just something I can take on Harvey in fifteen minutes. That’s not a certificate. That’s marketing.

Marlene Gebauer (32:40)
You raise a really good point about it takes time to solve these problems because I think that is also in itself a problem that these are not easy things to solve for a lot of times. And so but I think that the demand is that these are solved very quickly and immediately, and immediacy is really demanded. So again you know, talking to your clients, you know, how do you kinda prepare them for the fact that like, look, this could take six months to figure out and if you know, we gotta stay

Dr. Maryam Salehijam (33:18)
Yeah, I yeah, I’m

Marlene Gebauer (33:20)
the course.

Dr. Maryam Salehijam (33:20)
I’m super blunt. I tell them, I’m like, okay, you sound like you wanna Yeah. I’m like okay, sounds like

Marlene Gebauer (33:24)
No, you’re not.

Greg Lambert (33:25)
What? No.

Dr. Maryam Salehijam (33:27)
you wanna solve your matter management problem. How much time and money do you wanna spend on solving that problem? And is that actually a problem? Sounds like Excel’s working for you. Like, what would happen if you solved that problem? You’ll have more time for business development? Okay, cool. So it sounds like it’s worth it. So by the way it’s gonna cost sixty thousand and probably seven months of engineering if you wanna build yourself. So it’s really

Marlene Gebauer (33:53)
Right, right, right. And you’re the SME and so you have to be in it through the whole thing and you know, are you willing to do that?

Dr. Maryam Salehijam (33:59)
Because you want to build it. And then if you wanna buy it, guess what? You have to be willing to pay. You have to be willing to accept that the product sometimes sucks. You have to complain about your vendor like we all do, you know, but it’s more quick and after like a year or two you’re gonna say this solution didn’t work, we’re gonna replace it. I mean so typical. Everybody does this. Like nothing ever lasts, like other than LexisNexis or TR case law search. But like most things get pulled out and replaced and technology gets better and better product comes out. So again have to have that mindset of this is just the solution I’m using right now, and I might change my mind in six months to twelve months and do something else.

Greg Lambert (34:35)
Well what about and I found this description to be surprisingly effective when it comes to lawyers. when we talk about personal skill sets that you run into what’s called the J-curve, where when you implement a new technology or a change management that typically your skill set actually drops initially. before recovering and then going beyond where you were so it makes that J shape when it comes to improvement. do you use a similar argument to that to when you’re bringing in a new product, the change in the structure that it’s not going to be often up and to the right immediately. How do you how do you explain that to your customers?

Dr. Maryam Salehijam (35:25)
Yeah. Okay, this is again, I feel like I’m gonna get blacklisted because of all my opinions. I tell people, I’m like, if your people don’t know how to use a computer, I can’t teach them how to use AI. So it’s like talking about a J-curve, but I’m like

Marlene Gebauer (35:42)
Ha

Dr. Maryam Salehijam (35:43)
Actually like you want your people to get more efficient, just sign them up for a typing class. They go what? I’m like I just saw them share their screen and type. Literally a typing class is all they need. And they go what? I’m like, yep, three months typing class. That’s their typing is what’s slowing them down. And they go, Are you kidding me? I’m like, no. This is not an AI. It’s like they can’t copy paste. They can’t use the short like it’s literally sometimes that bad. And so I always walk it back to like, hey, let’s actually see what’s happening visibly. Look at it. And then we’ll know, The reason you’re spending 10 hours on a contract is because you don’t know how to use Control-F. Like this is And it hurts to hear it, right? Because everybody doesn’t want to like people don’t want to hear that they’re worse at something than other people. But literally I’ve seen lawyers not know how to use a computer. And it is concerning. And then they wanna go use AI, which is even more concerning because then they really don’t know what’s happening. And so that’s when we run into they submitted stuff that doesn’t make sense. They used it wrong. Because they don’t even understand how systems work. So that’s where I go again, wait a minute. Like, who’s going to be touching this? And who’s going to be approving this? Because you’re about to set yourself up for failure. If you give people access to AI engines who don’t know how systems and softwares and API calls work. So it’s again, education is a big part of my company to make sure people don’t put themselves in the hot seat.

Greg Lambert (37:11)
All right, well let’s say that you found the magic firm that has people that understands computers, can type, can use Control-F to find things on a on a

Dr. Maryam Salehijam (37:22)
Ha ha ha.

Greg Lambert (37:22)
document. but there’s still even an the next step. So we’re looking at bringing in AI tools and I think we’re still thinking of them almost as magical, you know, because we don’t quite understand them. But how do you get your get people to stop thinking of them as a as a like a separate tool and just start seeing it as part of just the infrastructure and the way that things get done going forward? How do you go to that level?

Dr. Maryam Salehijam (37:55)
Mm. That is a great question. And there’s like a lot of conversations in tech around unified interfaces. Basically you just have one system and the back end has all access points, right? And then there’s the conversations around B where people get their work done, which is why we see so many Word add-ins with lawyers. But to change the way somebody does work, like the way their brain goes and does the work, is really, really hard. And the only way you can actually do that is by repetition. So it’s about every day we’re gonna Have this many hours where we’re gonna all work together, right? Whether it’s sharing screens, being on one call, and we’re gonna have a pod lead that says, What are you working on? And they say this and this, and they go, Okay, I need you to try it this way, but they’re doing it in front of their peers. Why? Because peer pressure works. I know it sounds like everyone’s I don’t want to do that, but peer pressure works like magic. That’s why classroom settings work, that’s why asking people questions in front of the whole class works. So we need to bring back that methodology when we’re teaching people how to use some. something. Otherwise they’re gonna kinda listen, do something else, not really take it in. But if they have to prove themselves in front of their peers and we’re lawyers and we’re perfectionists, they’re gonna actually listen and do it and try to be good at it ’cause we all love to compete and show that we’re capable. So it brings back that law school setting into the AI department setting.

Greg Lambert (39:11)
I say we have to bring back the Socratic method to training.

Marlene Gebauer (39:13)
Socratic method I was saying the same thing.

Dr. Maryam Salehijam (39:16)
It works! Like why reinvent the wheel? We know how people learn, so why are we trying to do it differently?

Marlene Gebauer (39:23)
So Maryam, we’re going to ask sort of a question of looking at the past and then Greg’s gonna ask our crystal ball question about what’s happening for the future. So tell me what is true today that wasn’t true a year ago.

Dr. Maryam Salehijam (39:41)
Great question. What is true today that’s not true a year ago was that we thought we’re gonna see a lot more AI companies pop up in the legal space. A year ago we said that. I’m not seeing that. I’m not seeing too many new players. And I’m not sure if it’s because the venture capital firms don’t want to fund them or everyone’s watching the market. But I really thought there was gonna be an explosion of tools. But

Marlene Gebauer (40:03)
It’s interesting ’cause it’s like I see like a little but l little tiny like point solution types of things, but nothing big.

Dr. Maryam Salehijam (40:08)
Yeah. Nothing major. Nothing big, nothing major. and so that was surprising to me, ’cause I thought we would see a lot more of it. And maybe everyone’s in the build stage right now and we’re gonna hear about it in 2027. That’s true, yeah.

Marlene Gebauer (40:20)
And then there’s a lot of buying. There’s a lot of there’s a lot of acquisitions too. Yeah.

Dr. Maryam Salehijam (40:24)
So I think funding is drying up, actually instead of going up. And maybe it’s ’cause the frontier labs moved into legal. Maybe that’s what’s scaring the investors, but

Greg Lambert (40:33)
I was gonna say, ChatGPT just launched, what is it, Astra for Law? Yeah.

Marlene Gebauer (40:37)
Mm-hmm.

Dr. Maryam Salehijam (40:37)
Astra, yes. So that was surprising to me. I did not see funding dry up so quickly. For new players. Now, the crystal ball moment, what will happen? As a good lawyer, I’ll say it depends. It depends on the market, it depends on the environment, depends on global warfare. But if I was just to look six months ahead of from now, I would say we’re gonna see a lot of legal departments and law firms ask the question of can we build something we own that can do what the vendors do, but cheaper or better? We don’t have to renew it every year, right? We can maintain it ourselves, it’ll be our workflows, and we buy limited licenses to the vendors for the people that really need it. Because right now the way legal tech is getting sold is traditional SaaS, right? Seat-based pricing, minimum a year. Often they’re pushing three years. They want multiple licenses. But if you look at the usage data, it doesn’t even make sense. So I think we’re gonna see more of that shift coming because right now the spend on technology is not, back to the ROI conversation, supported by the ROI versus their price point.

Marlene Gebauer (41:53)
Well, Dr. Maryam Salehijam, founder and CEO of Andarzi, thank you for joining us on the Geek in Review.

Dr. Maryam Salehijam (41:59)
Thank you so much, Marlene. Thank you, Greg.

Greg Lambert (42:02)
And thanks everyone for listening and if you enjoy the show, please share it with a colleague. We’d love to hear from you on LinkedIn and Substack.

Marlene Gebauer (42:12)
And Maryam, what is the best place for listeners to learn more about Andarzi and follow your work?

Dr. Maryam Salehijam (42:18)
If they follow me on LinkedIn, I’ll post links to my website, andarzy.com, and also my courses on AI fluency via Maven.

Greg Lambert (42:26)
I feel like we’ve flipped the script here at the end, Marlene. So I’m taking your part. And

Marlene Gebauer (42:30)
We did. Like I noticed.

Greg Lambert (42:34)
and as always, the music you hear is from our good friend Jerry David DeCicca Thank you, Jerry.

Marlene Gebauer (42:39)
Thanks, Jerry.

Greg Lambert (42:40)
All right, bye everyone.

Marlene Gebauer (42:42)
Bye.

What happened when I pointed Jev, SALI, and CLOC’s Core 12 at every post we’ve ever written

I read Nate Jones’s post on Jev, TypeSafe’s new “System One” model, and he described it as an LLM that can only talk in multiple choice. It reads what you give it, you hand it the possible answers, and it picks one and tells you how sure it is. It can’t write you a sentence. My first reaction was, “What good is that?” My second reaction, about ten minutes later, was, “Wait... that’s what a cataloger does all day.”

So I did what any recovering law librarian would do with a new classification toy. I grabbed the biggest pile of legal-industry text I have the rights to, which is every post on 3 Geeks and a Law Blog going back to July 2008 (2,269 posts, about 3.9 million words, including 84 Geek in Review episodes with full transcripts), and I tried to catalog all of it twice. First against the SALI Legal Matter Specification Standard, then against CLOC’s Core 12 legal operations functions.

Here’s a link to the spreadsheet with the SALI and CLOC tags.

Let me walk you through how it went, warts and all.

The Setup

The division of labor is the whole trick, so let me explain it before the numbers.

Jev did the first pass. It’s cheap (about four cents per million input tokens, with output free) and fast (about a quarter of a second per call in this run). I called it through OpenRouter, since TypeSafe paused new signups the day I tried to get a key. Every call asks Jev a batch of yes/no questions about one post (“Is Knowledge Management a substantial topic of this post?”), and Jev returns a probability for each one.

Claude did the second pass, and it did it inside my regular Cowork session, which meant no per-call API charges. Any tag where Jev was unsure went into a review queue. Claude read the post and kept a tag only if it could quote a sentence, word for word, that supported it. A script then went back and checked every one of those quotes against the actual post text.

Code did everything else. It enforced the rules that a child tag can’t survive without its parent, capped each branch at five tags per post, and merged everything into one spreadsheet.

Before touching the blog, I also ran a warm-up on the podcast RSS feed (370 episodes). That’s where most of the early mistakes happened, and I’ll get to those.

Run 1: SALI

SALI’s standard is a family tree of 18,324 tags, published as an open ontology file on GitHub. I used four branches: Area of Law, Service, Document / Artifact, and Legal Use Cases (about 1,200 tags in all). Jev can only answer the questions it’s handed, so the script walks the tree like a game of 20 Questions. It asks about the 31 top-level Areas of Law in one call, then goes one level down only where Jev said yes, then one more level.

The calibration step. Before running everything, we pulled a 100-post sample spread across the years, including 25 transcript episodes, and had Claude review every tag Jev scored 0.30 or higher. That’s 1,610 judgments, made without Claude seeing Jev’s score. This is the part that saved the project, because it showed Jev’s confidence scores meant something different than I assumed:

On the podcast warm-up, I’d been letting Jev auto-keep anything at 0.70 or higher. On blog posts, that rule would have been wrong about half the time. The Document / Artifact branch was the worst of it. Below 0.80, Jev was almost always wrong about what kinds of documents a post discussed.

So we set a per-branch rule we called Policy A. Anything Jev scored at 0.90 or higher was kept automatically. A middle band went to Claude for review, and that band was different for each branch (0.50 to 0.89 for Legal Use Cases, 0.60 to 0.89 for Service, 0.80 to 0.89 for Documents, 0.30 to 0.89 for Area of Law). Everything else was dropped. The sample said that would give us about 94% precision while losing about 18% of the tags a full review would have found.

The full run, by the numbers:

• Jev made 241,498 yes/no decisions across 14,657 calls on all 2,269 posts. Total cost: about $3.35.
• Jev settled about 92% of those decisions on its own: 219,970 answers under 0.30 were dropped, and 1,536 answers at 0.90 or higher were kept (before the parent and cap rules trimmed that to 1,174).
• 6,336 decisions (about 2.6%) went to Claude for review, spread over 1,347 posts. Claude kept 2,118 of them, and 1,910 survived the parent and cap rules.
• Counting the calibration sample, Claude made 7,946 review calls on the SALI run, handled by 34 review agents running in batches.
• Final result: 3,431 SALI tags on 1,151 posts. Legal Use Cases carried the load with 2,569 tags, followed by Document / Artifact (443), Service (220), and Area of Law (199).

The most common tags were Business of Law, Legal Research, Pricing Management, Knowledge Management, and Legal Process Improvement, which sounds about right for a blog that has spent eighteen years arguing about billable hours and law libraries.

Run 2: CLOC Core 12

CLOC’s Core 12 is a flat list of twelve legal operations functions (Financial Management, Knowledge Management, Technology, Training & Development, and so on). With no tree to walk, each post got exactly one Jev call with twelve questions.

The calibration step used the same 100 posts, with 318 judgments this time. Jev behaved much better on the flat list:

We auto-kept everything at 0.90 or higher and reviewed 0.50 to 0.89. For Information Governance and Business Intelligence we reviewed down to 0.30, because the sample showed real tags hiding lower down. The expected loss was about 7%.

The full run, by the numbers:

• Jev made 27,228 yes/no decisions in 2,269 calls, in about two minutes, for about $0.39.
• 450 decisions were 0.90 or higher and kept automatically (403 after the sample posts were swapped for their reviewed versions). 22,682 were under 0.30 and dropped.
• 2,104 decisions (about 7.7%) went to Claude across 1,166 posts. Claude kept 1,298.
• Counting calibration, Claude made 2,422 review calls on the CLOC run, handled by 13 review agents.
• Final result: 1,849 CLOC tags on 1,190 posts. Technology led with 542, then Financial Management (214), Training & Development (149), and Knowledge Management (145).

What Worked

Letting Jev do the boring part. Across both runs, Jev answered 268,726 yes/no questions for under four dollars. Most of those answers were obvious “no” calls (a 2009 post about Twitter search tips is not about Bankruptcy Law), and Jev knocked them out at a quarter-second a call. That’s exactly the kind of job Nate described in his post, and on this project it held up well.

The quote rule. Requiring a verbatim quote for every kept tag turned out to be the single best quality control in the project. A script checked all of those quotes against the source text, and zero failed. It also means that every reviewed tag in the spreadsheet comes with its own receipt, which the librarian in me loves.

Calibrating before scaling. The 100-post sample cost thirty cents in Jev calls and one afternoon of review, and it kept me from shipping a dataset where a third of the tags were wrong.

Running Claude review inside Cowork. Moving the review out of the paid API and into my regular session changed the cost picture completely. The API version would have been something like $70 to $110 for the SALI run alone.

CLOC fit the content. A flat, twelve-item list turned out to be much easier for Jev than a deep tree, and it maps better to what this blog is actually about. Knowledge Management shows up on 145 posts under CLOC, compared with 83 under SALI.

Transcripts held up. I expected the long, rambling podcast conversations to confuse things. They didn’t. Transcript posts and regular posts scored about the same (55% and 53% precision at 0.70 and up in the SALI sample).

What Was Questionable

SALI is the wrong shape for this corpus. SALI was built to describe legal matters. Area of Law barely registered (199 tags across 18 years), and most of the work landed in Legal Use Cases. That’s a reasonable outcome for a business-of-law blog, but it tells me SALI is a better fit for matter data and client documents than for commentary about the industry.

Nearly half the posts have no SALI tags. 1,118 posts came out with no SALI tags, and 805 have no tags from either taxonomy. Plenty of those deserve to be empty (I’m not sure “Merry Christmas!” or “The Leadership and Management Styles of Dolly Parton” needs a SALI tag), but some are real misses. The AALL 2026 Annual Meeting preview should have picked up Legal Research, and it didn’t.

The reviewers haven’t been reviewed. All 10,368 review calls were made by Claude. Nobody has graded Claude against a human yet, and that includes me. The quote rule tells me the evidence really is in the post, and I still need a person to tell me whether Claude read it the right way.

The 84% auto-keep rate on SALI. Tags Jev scored 0.90 or higher went straight in without review. On the sample that was right 84% of the time, so something like one in six of those 1,174 auto-kept SALI tags is probably wrong.

Jev’s confident mistakes. Jev said a confident “yes” to Expert Opinion and Template on the Best Lawyers episode, and to Opinion Memo Practice on the LexisNexis CTO interview. That’s why the per-branch thresholds exist. Nate made this point in his post as well: calibration is measured across thousands of answers, so a 0.95 on any single post is still just a strong hunch.

What happened below 0.30. We never reviewed anything Jev scored under 0.30, so whatever real tags are hiding down there are unmeasured.

What Failed

The first dollar. The first pilot sent Jev’s close calls to Claude through the paid API. Claude kept all 85 of the borderline tags it reviewed, which makes a review step pretty pointless. Then 17 episodes failed after Claude had already answered and been billed. Fifteen of those failures happened because a fix I thought had been saved to my computer never actually landed, so the old script kept running. To top it off, the script didn’t count the cost of failed episodes, so my $1 key limit was gone before the report said it should be. We fixed the logging, tightened the prompt, started checking every saved file with a checksum, and eventually moved the review into Cowork.

The RSS feed was the wrong source. The podcast feed cuts show notes off at about 4,000 characters, right where the “Transcript” heading starts. Everything on the first podcast run was tagged from summaries only. The blog feed has the full text, which is why we switched.

I knocked over my own blog. While downloading the archive, about 65 pages in, geeklawblog.com started throwing Cloudflare 520 and 521 errors. My back-to-back requests probably didn’t help. We waited for it to come back and pulled the rest slowly, one page every few seconds.

Sponsor reads. About half the recent transcripts open with a Legal Technology Hub segment on AI governance, which would have tagged every episode with AI governance. The first attempt to strip it out missed most of them. A second pass that looked for the sponsor’s guest speaker did better, but it’s still a rough heuristic.

Where This Leaves Me

For less than five dollars in Jev and OpenRouter charges, plus about 47 review agents’ worth of Cowork usage, every post on the blog now has two sets of standard tags. Every tag comes with a Jev score, a record of who made the call, and in most cases a quote from the post itself. That’s a catalog I would have killed for back when I was running a law firm library with a card catalog and a budget line for “cataloging backlog.”

The part that sticks with me is how much the calibration step mattered. I went in assuming Jev’s confidence scores could be taken at face value, and on a deep taxonomy like SALI that assumption would have burned me. Thirty cents and 100 posts worth of spot-checking showed me which score ranges needed a second reader.

This was a fun initial test that presented a few areas that need improvement, but there is a lot of potential here to open up a number of decision processes at a very affordable price.

Paul Lee, co-founder and CEO of Patlytics, joins Greg Lambert to explain how an AI platform built specifically for intellectual property work is changing the patent lifecycle. Patlytics supports workflows spanning patent drafting, prior art analysis, office action responses, portfolio management, litigation readiness, and claim-chart preparation. Lee reports that the company now works with roughly 55 percent of the Am Law 100 and hundreds of corporations across technology, biotechnology, pharmaceuticals, and other patent-intensive industries.

Lee traces Patlytics’ origins to his experience as a venture capitalist and more than 100 conversations with patent attorneys. Those interviews exposed a practice filled with expensive, labor-intensive processes, from drafting detailed patent specifications to constructing claim charts for litigation. His interest also grew from the Apple and Samsung patent battles, the IP expenses faced by venture-backed companies, and conversations with Patlytics co-founder Arthur Jen and former Latham & Watkins patent litigator Bob Steinberg.

The conversation turns to Patlytics’ work involving USPTO patent examiners and the broader effect of placing AI on both sides of the examination process. While confidentiality limits the details Lee discusses, he identifies quality and the examination backlog as two areas where specialized technology offers meaningful assistance. He also contrasts Patlytics with broad legal AI platforms such as Harvey and Legora, arguing that patent professionals need tools designed for the precision, technical detail, and specialized workflows of IP practice.

Human judgment stays central to Lee’s vision. Patent attorneys still own the work product, approve key decisions, and remain responsible when an AI-generated analysis falls short. At the same time, client expectations continue to rise. Clients want faster work, higher quality, and lower costs, while law firms need sustainable margins. Lee sees flat-fee arrangements and more predictable workflows as one route toward sharing the “AI dividend” between clients and their outside counsel. In-house teams also gain more capacity for infringement analysis, patent-portfolio reviews during M&A, cross-licensing strategy, and litigation preparation.

Looking ahead, Lee describes a striking change in attitude among patent professionals, from widespread skepticism a year ago to broad optimism today. His crystal-ball concern is less about whether lawyers will adopt AI and more about whether its economics will hold together. As free experimentation gives way to consumption-based pricing, firms will need to measure the value of each workflow and avoid spending $50,000 in AI costs on a $5,000 matter. Token maxing had its moment. ROI gets the next meeting invitation.

Listen on mobile platforms:  ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Apple Podcasts⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ |  ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Spotify⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ | ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠YouTube⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ | ⁠Substack⁠

[Special Thanks to ⁠⁠Legal Technology Hub⁠⁠ for their sponsoring this episode.]

Email: geekinreviewpodcast@gmail.com

Music: Jerry David DeCicca

Transcript

Greg Lambert (00:00)

Hey, everybody. This week we have Paul Lee from Patlytics on the show, and we’ll be talking about patents, AI, the USPTO, and a whole lot more. But first up, here’s a word from our sponsor.

Marlene Gebauer (00:19)

Hi, I’m Marlene Gebauer from The Geek in Review. I have Sam Moore here from Legal Technology Hub. He’s going to tell us about AI governance from the advisory side.

Sam Moore (00:28)

Thank you. Good to see you again.

In the advisory team at Legal Technology Hub, we’ve been having a lot of conversations with clients about AI governance. Most law firms and departments have at least an AI policy of some kind, and there are some broad standards starting to emerge.

A trend we’re seeing at the moment is a shift away from policies written in 2023 and 2024, which largely stated what you couldn’t do, toward a more informative approach that defines different risk categories and makes distinctions between routine, low-risk uses of AI and other, more substantial use cases.

I think that’s appropriate for where we are. It’s also what clients are coming to expect from their advisors.

One of the bigger challenges we’re seeing right now is how you express an AI use policy in such a way that any member of your legal team could explain it to a client. I think that’s still a big challenge for the industry.

I don’t think clients are all that happy if their attorney says, “I’ll have to get back to you about that,” because the attorney is likely using AI day to day. They should be able to explain the policy themselves. How else do you know they’re complying with it?

We’re doing some interesting work at the moment helping law firms and law departments take those first-version AI use policies, which are often very much “thou shalt not,” and turn them into more readable, appropriate governance positions that inform the conversation with clients.

If anyone wants to find out more about our advisory services, they can find me or Cheryl Wilson Griffin on LinkedIn, or visit legaltechnologyhub.com.

Marlene Gebauer (02:09)

Thank you, Sam. It is important to have an AI policy in plain language so people understand it.

Sam Moore (02:16)

Absolutely.

Greg Lambert (02:17)

Welcome to The Geek in Review, the podcast focused on innovative and creative ideas in the legal industry. I’m Greg Lambert, flying solo this week. Marlene is out.

This week I am joined by Paul Lee. Paul’s the co-founder of Patlytics. Paul came to legal technology from the venture capital side, and he and his team have built Patlytics specifically around patent workflows, from drafting and prior art analysis to office actions and litigation.

Paul, welcome to The Geek in Review. Good to have you here.

Paul Lee (02:52)

Hey, Greg. Thanks for the invite. Awesome to be here.

Greg Lambert (02:55)

Paul, before we jump into the questions, I wanted to see if you would give us a quick elevator pitch on what Patlytics does and what problems your customers come to you to solve.

Paul Lee (03:09)

Yeah, absolutely.

Patlytics’ mission since day one, and that hasn’t changed since founding the company, has always been to create exceptional value in the IP ecosystem.

The way we do that is to transform and reduce transactional inefficiency, from the creation of IP and patents, from drafting to office actions on the prosecution side, to portfolio management of IP and patents for corporations and the law firms we work with, to litigation readiness and litigation itself.

The whole lifecycle is something I saw as an opportunity where we could have a massive impact. I think we’ve done right by our customers one by one by making sure we help as much as possible.

Greg Lambert (04:12)

Great. Thank you for doing that.

Let’s jump in. Paul, before building Patlytics, you took the time to learn the industry. I think you spoke with more than 100 IP professionals about how patent work gets done.

Having worked with my IP lawyers, I understand it’s highly detailed work. Sometimes it is low-margin work, so efficiencies built into the process are likely welcome by the community.

When you were having those conversations, what did they reveal about patent practice that convinced you Patlytics would be a valuable tool?

Paul Lee (04:58)

Yeah, that’s a great question, Greg.

It’s funny to think about where we are now and where we were when it was only an idea and we were starting up.

Right now, we work with around 55 percent of the Am Law 100 firms. We work with hundreds of corporations, from massive biotech and pharma companies to technology companies across Asia, North America, Europe, and elsewhere.

To understand the momentum we have now, you need to go back to where we were when things were starting.

For background, I do not have a patent attorney background. I was a venture capitalist, as you mentioned. One thing you do learn in venture capital is how to talk with a lot of people and potentially see patterns and insights.

In the early days, while figuring out whether this was an interesting opportunity, I spoke with more than 100 patent attorneys. Confidently, I would say many of them told me at the time that a lot of the processes were quite inefficient.

The fact that you might spend more than 10 hours writing a detailed specification for an application was wild.

Creating a claim chart used for IP litigation, with the level of effort, scrutiny, and back-and-forth involved, was something I saw firsthand. I thought that was pretty wild as well.

When you compound all of these use cases and processes across different parts of the IP lifecycle, I thought there would be tremendous value if we figured out how to create one platform to help with each of those workflows.

Greg Lambert (07:09)

I’m curious, did this idea come to you because you were working around patents, or how did you stumble across patents as a particular industry that needed this?

Paul Lee (07:24)

I first heard about patents back when I was a student. I was in Waterloo, and there was that massive Apple versus Samsung patent litigation battle happening.

At the time, as a student, I was shocked by the dollar value of all that.

Fast forward a couple of years. As a venture capitalist, you tend to invest in a lot of IP companies. Some of those companies spend a lot of money on trade secrets, patents, trademarks, and so on.

From my seat in venture capital, I was curious about how the bill could get so high on certain IP litigation matters. That got me curious.

My co-founder and CTO, Arthur, while he was building Magic, dealt with patent litigation and patent infringement matters.

Then I think the idea took off when my friend Bob Steinberg, who was chair of the IP litigation practice at Latham & Watkins, invited me over to his place and we discussed where he saw the opportunities.

It was a trifecta. It wasn’t one apple-hitting-the-ground kind of moment. The topic kept coming up, and that’s how we got started.

Greg Lambert (09:03)

I know a lot of startups are told to build into their budget the possibility of patent litigation or IP litigation at some point because many of them will face that.

I want to talk about something everyone I’ve talked to about Patlytics mentions, the fact that the USPTO is using Patlytics for its patent examiners.

What changes about the patent system when AI becomes part of the workflow on both the applicant side and the examiner side? What’s AI bringing into the overall workflow?

Paul Lee (09:50)

Greg, I need to preface this with the hundreds of pages I had to sign around confidentiality and the details of what we may or may not do with government agencies.

There are public records out there that people can search, but for this podcast, let me stay pretty high level about what we aim to do.

Greg Lambert (10:19)

Absolutely.

Paul Lee (10:20)

Going back to my original point, the patent and IP ecosystem is highly laborious and transactionally inefficient.

A good chunk of that exists at the corporate and law firm side because there are a lot of use cases, a lot of work, less budget, and the need to do more with less.

From the examination and government side, historically there have been opportunities and room for improvement as well.

We believe the right technology has the opportunity to add a lot of value to the whole flow. It has been rewarding for us to work with Google to make sure we’re on the right footing with everything we’re doing with government agencies.

What I would say is that USPTO examiners do a lot, and there’s a backlog associated with the amount of work involved.

The opportunity we have is to help make sure things remain high quality while helping with the backlog as much as possible.

Greg Lambert (11:56)

I want to talk about the way people are looking at AI through these broad AI platforms, the Harveys, Legoras, Westlaw, Lexis, and others.

I want to talk about being a specialist versus being a platform. What does a specialized patent platform like yours need to do to stand out and earn a permanent place in the law firm’s technology stack? What makes Patlytics shine?

Paul Lee (12:40)

It’s something I think about all the time, and something all of us as a team spend a lot of time thinking about.

The answer is relevance, Greg.

The question is whether we’re earning the right to be 10 times better than the status quo and provide a much better experience than generalized legal tools or the previous way of doing things.

That’s always the benchmark. The benchmark is what people did previously.

For us, we want to make sure the experience and outcomes derived from Patlytics are delightful and 10 times better.

I have a lot of respect for Harvey, Legora, and some of the names you mentioned. There’s always going to be a place for them. Some of the most interesting transformation in law is happening right now because of players like that.

With that said, what I hear over and over again is that the IP practice felt unloved.

Greg Lambert (14:03)

It’s those scientists. They never feel loved.

Paul Lee (14:08)

Exactly.

The IP practice feels the generalized tools, even with the ability to create your own workflows and make things more customized, never fully get there. They get to 80 or 85 percent.

We think of the patent ecosystem like an F1 race. It’s about getting to 99 percent.

When we started the company, I thought getting to 85 or 90 percent would be enough. That’s no longer the benchmark, and the benchmark moves year over year.

From a product perspective, we think about this and act on it constantly.

Compared with the existing flow for creating charts, creating applications, handling office actions, or pruning a portfolio, is the experience we’re putting in front of our customers genuinely 10 times more valuable than the previous way of doing things?

Greg Lambert (15:23)

We joke about scientists not feeling loved, but there are a lot of technical processes involved in setting up a patent filing.

You are moving into work such as patent drafting, office action responses, claim analysis, and even patent figures.

Where do you think technology should deliberately stop and require the patent lawyer to exercise judgment?

Or, even more broadly, where does it need to stop so the USPTO doesn’t look at the work and say AI did so much of this that it’s no longer eligible for a patent? What are you hearing in those areas?

Paul Lee (16:23)

I’m hearing a lot of things, but if I were to pinpoint one or two topics that keep coming up, number one is the demand to make sure human judgment is placed in the right spots.

At the end of the day, the practitioner still signs off, gives approval, and owns the outcome of the work product.

At Patlytics, we fundamentally believe there will always be a human-computer symbiosis. The human part will focus on judgment, while the AI and computer side helps develop the most useful ways to deliver those outcomes.

But the workflow will always require human judgment at the most important points.

That’s why tools in the past that said, “You don’t need to lift a finger. Click a button and the application gets submitted,” never worked out.

There are similar examples where people say, “Type something in, it drafts a claim chart, and you submit it.”

In fact, there has been some recent evidence of people being pulled into malpractice matters because of cases like that in the patent system.

We’re diligent and careful about this. It starts with the philosophy of human-computer symbiosis and making sure there are appropriate checkpoints across workflows, from prosecution to portfolio management to litigation.

I would add that client expectations are becoming higher as well.

Maybe it’s because of AI, or perhaps something else, but the client expectation is, “Use AI, but make sure the result is as good as it would have been if you spent many hours on it. We also expect it to be done in a fraction of the time, with even higher quality.”

That level of expectation puts law firms, and IP attorneys especially, under enormous pressure.

Going back to what you said, in intellectual property and patents specifically, each word matters so much.

Greg Lambert (19:31)

Absolutely.

Paul Lee (19:31)

I know in law each word matters. There are redlines and all of that.

But from patent drafting to patent litigation, this has given me a new sense of respect for the type of work patent attorneys do.

Greg Lambert (19:50)

Pulling on the thread of client demands, one of the biggest topics around AI in legal practice is who gets the AI dividend at the end of the day.

If AI turns work that once cost thousands of dollars in attorney time into something much faster, who ultimately captures the economic value of that AI dividend?

Paul Lee (20:31)

That’s a fascinating point, Greg.

I don’t think the market has figured that piece out yet. We’re moving through a lot of exploration and back-and-forth around this as we speak.

What I would say with certainty is that the transactional inefficiency of the past has come down.

Clients want collaboration. Law firms want faster collaboration done the right way. A lot of that has been achieved through what legal AI has become over the past couple of years.

When it comes to the economics of the value chain, clients want to capture 100 percent of that efficiency. Law firms want to capture as much of those efficiencies as possible as well.

I think the answer is meeting somewhere in the middle.

If the business model of law changes in a way where both sides benefit tremendously, that’s great, and there are examples of that happening right now.

We work with many Am Law 100 firms. One firm we work with is in the top five and historically was known for rarely using flat-rate arrangements.

Greg Lambert (22:10)

Right.

Paul Lee (22:12)

With AI, IP, and the way Patlytics is serving its offices and clients across these practice areas, an interesting business model has emerged.

On certain matters, we’ve seen more flat-rate business models where lawyers benefit because they are better able to predict how long something will take. Margins become more predictable, and lawyers are able to cycle through matters faster when those efficiencies are in place and demand exists.

On the other side, corporations are getting what they have wanted all along, which is greater predictability and lower bills.

I’d call that a win-win situation, and I think we’ll continue to see more of it.

Greg Lambert (23:11)

We talked briefly earlier about Patlytics being used both in-house and by outside counsel.

Where are you seeing that division of labor going? How much more upfront work is the in-house team doing before turning it over, and how has the division of who does what changed when it comes to patents?

Paul Lee (23:47)

For us at Patlytics, everything started with law firms in the first year. That’s where the demand was.

Take the patent prosecution practice as an example.

Historically, it has been one area where costs kept rising over the past 10 years, including inflation, while pricing hasn’t changed much. If anything, pricing has gone down a little.

You have costs going up on one side while the top line comes down. That leaves extremely thin margins. It’s a tough business.

What we’ve been able to accomplish is lowering the cost component in terms of hours and the way work gets done.

Even if the flat rate goes down a little, the margin calculation becomes much better for the prosecution practices we work with.

I’ve had partners at law firms tell me, “I had a record year because of Patlytics and the tooling we’ve put in place.”

I don’t want to get into specific numbers, but this particular partner was extremely happy with the outcome in his practice.

To your question about corporations and law firms and what work gets done where, corporations now have the ability to do more with less. They want to do things they simply didn’t have time for in the past.

I think Patlytics has helped in-house counsel do things such as running infringement analysis before a continuation, looking at the market more quickly before deciding which actions to take in a patent strategy, assessing patent pool M&A opportunities, and having greater confidence about where they’re headed for litigation readiness or potential cross-licensing.

They’re able to get down to claim construction and element-by-element analysis and, eventually, perhaps have greater insight ahead of a Markman hearing.

Greg Lambert (26:30)

Mm-hmm.

Paul Lee (26:31)

Those are more sophisticated outcomes that in-house counsel simply didn’t have much time for before.

We haven’t even talked about R&D, inventions, and engineering.

One of the biggest pain points for patent attorneys is dealing with many engineers submitting material. Attorneys have to parse through a lot of noise to find the signal.

That’s a laborious process.

We saw tremendous value in bringing all of these use cases, or “jobs to be done,” under one umbrella to provide a one-plus-one-equals-three type of experience.

Greg Lambert (27:23)

The more I’ve gotten into patents over the years, the more I realize it isn’t a straight line. There are so many things going back and forth.

Paul Lee (27:35)

Yeah. It’s like drawing many different circles, then drawing 20 zigzag lines, and saying, “Hey, figure it out.”

Greg Lambert (27:45)

An economist’s chart dream right there.

All right. A lot has changed, but is there something true for you, Patlytics, or the industry today that wasn’t true a year ago?

Paul Lee (28:05)

Greg, I’ll talk about patent attorneys and in-house attorneys who deal with IP.

A year ago, the sentiment was perhaps 20 percent optimists and 80 percent still somewhat skeptical.

Now I’d say it’s 95 percent optimists and 5 percent skeptical.

Within one year, we’ve gone from conversations where a firm partner or managing director said, “We’re not going to look into this for the next five or 10 years, if at all,” to many instances where I get a call saying, “Scratch what I said. Could we start something tomorrow?”

This is truly one of the most interesting times in legal. There’s a transformation happening that I think is generational and will leave a lot of second-order and third-order consequences.

Because the whole ecosystem is becoming AI-forward, people are now asking, “What else should we do?”

Greg Lambert (29:29)

Yeah.

Paul Lee (29:30)

There’s much more creativity, more pushing of limits, and more willingness to take bigger shots.

Greg Lambert (29:36)

That leads perfectly into my last question.

We ask all of our guests to pull out their crystal ball and peer into the future for us.

You’ve taken us from last year to this year. What challenges or changes do you think the industry is going to face that we should prepare for over the next two to three years?

Paul Lee (30:05)

That’s a tough one.

Greg Lambert (30:06)

Predictions are hard, especially when they’re about the future.

Paul Lee (30:10)

Yeah. I hope I’m right, because five years from now I want to look back at this.

Greg Lambert (30:15)

We’ll bring you back in a year and see how well you’re doing.

Paul Lee (30:19)

Greg, I’d say there’s a lot of buzz around AI right now and the use of AI, sometimes without enough intention or purpose.

I think in the future its use will become much more intentional.

One thing I deal with right now is creating a layer of transparency so costs don’t go through the roof from an LLM perspective, both for suppliers and for us as a customer of the main suppliers fueling the entire system, the LLMs.

I don’t think the level of growth we’ve seen around LLMs is going to change. People will keep using them and become more intentional about their use, while making sure AI continues to prove value.

But the economics of AI, from the LLM to the end customer, are where I expect to see even more change.

From the venture side, the market has often viewed AI as great because, when you invest in a company, the more people use the product, the more revenue the company generates.

The flaw in that logic is that customers have a budget for this kind of thing. There’s always a budget.

Even if your business model depends on consumption continuing to grow, a budget is a budget. There is only so much a customer will spend.

I think we’re going to deal with that type of correction pretty soon.

It will increasingly come down to proving value and ROI in a meaningful way.

Greg Lambert (32:14)

I think we’ve come to the end of our free ride when it comes to LLMs, and now we’re looking at consumption-based pricing.

You want to make sure you’re not spending $50,000 pursuing a $5,000 matter.

I imagine on the business side you don’t have a “max token” scoreboard anymore. It’s more about being smart about how you’re using the technology.

Paul Lee (32:47)

Yeah. Token-maxing is likely not going to be a thing anymore, unless the LLM economics make sense for the whole value chain.

Greg Lambert (32:57)

I’m waiting for it to go the way of text messaging.

We’ll pay 10 cents a message for a couple of years, then hopefully you buy one plan and we go back to unlimited.

Paul Lee (33:06)

I remember that.

Being a Waterloo guy, that’s why BBM was so awesome. Did you have a BlackBerry back in the day?

Greg Lambert (33:13)

I did not.

Somehow I missed that and went straight from my Palm Pre to an iPhone.

Paul Lee (33:25)

That’s even better.

Greg Lambert (33:29)

Paul Lee from Patlytics, it’s been a pleasure having you on here and geeking out. I appreciate you being here.

Paul Lee (33:38)

Yeah, I had a lot of fun, Greg. Thanks for the invite.

Greg Lambert (33:41)

Absolutely.

And thanks to everyone who’s listening. If you enjoyed the show, please share it with a colleague. I’d love to hear from you on LinkedIn or Substack.

Paul, where’s the best place for people to learn more about Patlytics or the things you’re up to?

Paul Lee (33:59)

Definitely LinkedIn. Feel free to add me.

I’m in conversation with a lot of forward-thinking IP leaders and innovation folks from corporations and law firms, and I’d love to chat.

Also, feel free to give us a shout through the Patlytics homepage at patlytics.ai.

Greg Lambert (34:28)

All right. Thanks again.

And as always, the music you hear is from our friend Jerry David DeCicca.

Thank you, and talk to you later.

Paul Lee (34:37)

See you, Greg.

What happens when you separate elite legal talent from the traditional law firm structure? This week on The Geek in Review, we talk with Manuel Deó, co-founder and co-CEO of Ambar Partners, about a model designed around senior independent lawyers, flexible capacity, enterprise technology, and client choice. Deó explains why he and co-founder Rosa Espín did not set out to replace Big Law, but instead to address a gap between permanent in-house hiring and traditional outside counsel.

At the center of Ambar’s model is a simple idea: legal demand comes in different shapes, and the delivery model should match the problem. Deó describes how Ambar gives lawyers control over the clients, projects, fees, and schedules they take on, while giving clients greater visibility into cost and the individual lawyers doing the work. He also discusses Ambar’s recent Chambers recognition and argues that the term “alternative” is starting to lose some of its usefulness as clients grow more comfortable assembling legal services from a wider range of providers.

Deó walks through what Ambar calls its legal operating system, built around belonging, business, backbone, and badge. The model combines a professional community with business development, compliance, contracting, billing, technology, and institutional credibility for independent lawyers and specialist boutiques. Ambar’s public materials describe a shared technology environment that includes tools such as Harvey, Microsoft Copilot, Legora, and other legal technology products. The goal, according to Deó, is to give independent lawyers access to the infrastructure associated with a large firm without requiring them to give up professional independence.

The conversation also turns to AI, knowledge, and professional judgment. Deó argues that legal knowledge is becoming more abundant while judgment grows more valuable. He describes Ambar’s work on “expert twins,” where approved knowledge, prior work, playbooks, and experience associated with an individual lawyer form a trusted layer for AI-assisted work. That emphasis on institutional knowledge and permissions echoes a broader trend across legal AI, where vendors are increasingly focused on connecting AI systems to trusted internal work product and organizational context.

Finally, Deó offers a broader view of where legal delivery is heading. He sees legal departments assembling teams dynamically from in-house lawyers, traditional firms, independent specialists, boutiques, managed services, and AI agents based on the needs of a particular matter. Instead of asking which firm to hire, clients increasingly have reason to ask what combination of people, technology, expertise, and risk structure best fits the work. For Deó, the future belongs less to a single dominant delivery model and more to legal departments acting as orchestrators of capability.

Listen on mobile platforms:  ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Apple Podcasts⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ |  ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Spotify⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ | ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠YouTube⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ | ⁠Substack⁠

[Special Thanks to ⁠⁠Legal Technology Hub⁠⁠ for their sponsoring this episode.]

Email: geekinreviewpodcast@gmail.com

Music: Jerry David DeCicca

LINKS

• Ambar Partners
• Ambar Community
• Ambar Hauss Membership Plans
• Dr. No Newsletter

Transcript

Greg Lambert (00:01)

Hey everyone, I’m Greg Lambert, and this week on The Geek in Review, we have with us Manuel Deó from Ambar Partners. But first, let’s get some insights from our friends at Legal Technology Hub.

Marlene Gebauer (00:14)

Hi, I’m Marlene Gebauer from The Geek in Review. I have Sam Moore here from Legal Technology Hub. He’s going to tell us about AI governance from the advisory side.

Sam Moore (00:23)

Thank you. Good to see you again. Well, in the advisory team at Legal Technology Hub, we’ve been having a lot of conversations with clients about AI governance. Now, most law firms and departments have at least an AI policy of some kind, and there are some broad standards which are starting to emerge. A trend we’re seeing at the moment is a shift away from policies written in 2023 and 2024, which largely stated what you couldn’t do, and towards a more informative approach which defines different risk categories and makes distinctions between routine, low-risk use of AI and other more substantial use cases. I think that’s appropriate for where we are. It’s also what clients are coming to expect from their advisors. But one of the bigger challenges we’re seeing right now is, how do you express an AI use policy in such a way that any member of your legal team could explain it to a client? I think that’s still a big challenge for the industry.

Marlene Gebauer (00:41)

Okay.

Sam Moore (01:21)

I don’t think clients are all that happy if their attorney says, “I’ll have to get back to you about that,” because the attorney is using AI day-to-day. So they should be able to explain the policy themselves. How else do you know they’re complying with it? We’re doing some interesting work at the moment helping law firms and law departments take those first-version AI use policies that are very much “thou shalt not” and turn them into more easily readable, appropriate governance positions that inform the conversation with clients. If anyone wants to find out more about our advisory services, they can find me or Cheryl Wilson Griffin on LinkedIn, or they can visit legaltechnologyhub.com.

Marlene Gebauer (02:05)

Thank you, Sam. I mean, it is important to have AI policy in plain language so people understand it.

Sam Moore (02:11)

Absolutely.

Marlene Gebauer (02:19)

Welcome to The Geek in Review, the podcast focused on innovative and creative ideas in the legal industry. I’m Marlene Gebauer.

Greg Lambert (02:26)

And I’m Greg Lambert, and today we are examining the complete deconstruction of the traditional law firm. For decades, the legal industry was defined by centralization, kind of bringing the talent, the technology, the overhead, and the clients under one structured, highly leveraged roof and partnership pyramid. But a new model is proving you can unbundle law firms, keeping the elite talent and the enterprise technology while stripping away the rigid hierarchy.

Marlene Gebauer (02:56)

Our guest today is operating at the absolute forefront of this shift in the European and Latin American markets. Manuel Deó is the co-founder and co-CEO of Ambar Partners, which recently secured a Band One ranking in Chambers as a leading alternative legal service provider. Before launching Ambar, Manuel was the youngest Spanish lawyer to make partner at Latham & Watkins, eventually leading their corporate and M&A practice in Spain.

Greg Lambert (03:25)

And Manuel and his co-founder, Rosa Espín, have built what they describe as a legal operating system. It’s a digital platform called Ambar Haus that connects a highly vetted, curated community of over 400 senior independent lawyers with multinational enterprise clients. They are deploying enterprise tools like Harvey AI at scale, across borders, and completely rethinking how legal capacity is consumed and built.

Marlene Gebauer (03:55)

Manuel, welcome to The Geek in Review.

Manuel Deó (03:57)

Thank you, Marlene. Thank you, Greg, for having me. It’s a true pleasure. Thank you.

Marlene Gebauer (04:03)

So Manuel, you went from leading corporate and private equity at Latham & Watkins in Spain to co-founding Ambar. What did you see in the traditional legal market that convinced you to build a different kind of legal business?

Manuel Deó (04:18)

First, I should say that I loved Big Law, right? I always consider myself a son of the old model, and I’m a grateful son for many different reasons. Latham, in particular, where I was a partner, taught me what excellence, intensity, and global execution look like.

I did not build Ambar because I believed traditional law firms were broken. This is something people sometimes ask me. I built it with Rosa because no single institution is optimized for every type of legal need. We kept seeing two false choices when we were back at Big Law.

On one side, you had outstanding lawyers who had to choose between institutional prestige and professional autonomy. On the other side, you had clients who had to choose between adding permanent headcount to their teams or engaging a traditional law firm, usually through a narrow model, an hourly-rate basis.

But a significant part of corporate legal demand is neither permanent nor junior. It is senior, important, urgent, and variable. The market has never been short of talent. I think quite the opposite. What it’s been short of is an architecture capable of deploying that talent differently.

We wanted to separate expertise from real estate in specific offices on specific streets, seniority from hierarchy, and value from time. I think the traditional legal market had two gears. One is hire a lawyer, or hire a law firm. What we built is a gearbox. I think that’s the broader opportunity we’re building here. Not to replace law firms or internal legal departments, but to give clients a more intelligent, flexible way of building legal capacity around the problem they have.

So that’s what we’re doing and why we went from Big Law to building Ambar.

Greg Lambert (06:38)

As you were building this with Rosa, how were you thinking of structuring it in a way that was both enticing for the lawyers you wanted to recruit and also for the clients to understand what the business model was and what you were offering?

Manuel Deó (07:02)

Right. I think for lawyers the most important part of our model is freedom. We give lawyers freedom. Freedom to decide which projects they are engaged on, for which clients, for what fees, on what terms, and at what moment of the year they want to work. They have full control to decide which type of projects they are a part of. That’s the main value proposition for lawyers.

On the other side, for clients, it’s twofold. One is having full control. Full control over the cost, and also over the specific lawyer delivering the legal service. When you’re part of other, more traditional models, you hire a firm, but you don’t always have full control over who’s doing the job, right? Who’s doing the legal work? Because they have a number of people there doing the legal delivery.

So from one specific point, that’s full control over cost and delivery. And second, access to highly vetted lawyers where you have full control over the talent, expertise, and knowledge of the lawyer, who is highly vetted by third parties, not only by us, but by other clients, other colleagues, and, obviously, traditional firms.

Marlene Gebauer (08:17)

See, this is interesting. When we talk about value, we’re usually talking about value to the client, but I think you’re talking about it going both ways. What does the lawyer find valuable to work on, as well as delivering value to the client?

Manuel Deó (08:29)

Yeah, absolutely.

Greg Lambert (08:32)

Marlene mentioned this in the intro, but congratulations on Ambar’s recent Chambers recognition as Band One for ALSPs in Europe. Both you and Rosa were also named individually in Band One as well. The Chambers submissions, I know outside of legal no one knows what those are, but it’s a big deal inside legal. So what does that recognition tell you about the legal market’s acceptance of the type of model you’re providing?

Manuel Deó (09:07)

Right. The recognition is specifically for what they call NewLaw, or alternative legal service firms. What that tells me is that the word “alternative,” which is a buzzword used in the sector for many years, is approaching its expiration date.

We’re grateful for a prestigious institution like Chambers recognizing Ambar as Band One with other global companies, but clients validated the model before Chambers did. Chambers has made visible something that was already happening.

I think the true thing is that sophisticated clients are increasingly indifferent to labels and brands. They care about quality, speed, accountability, commercial understanding, and value. The general counsel, which is the name given to the most senior person in the legal department at big corporations and S&P 500 companies, are sophisticated people with a lot of pressure in their companies. They’re typically trusted advisors to the CEO and have a lot of pressure to deliver more value with less.

This recognition demonstrates that top-tier standards exist outside the traditional partnership pyramid. Independence, which is part of our value for lawyers because they are independent lawyers, does not have to mean fragmentation. Flexibility, which is something we provide to our clients, does not have to mean lower quality. Quite the opposite.

A lawyer does not have to sit inside a large institution for that lawyer’s expertise to be institutional-grade, relevant, and appealing to a big client. We are proud of that recognition. But the ranking is a mirror, it’s not a moat. The question is no longer whether clients will accept these models. I think they already do.

The hard question for me is who will scale the model without diluting trust, quality, or accountability.

Marlene Gebauer (11:33)

So I want to dig in a little bit at how the platform works.

Manuel Deó (11:38)

Sure.

Marlene Gebauer (11:39)

Ambar describes itself as a global platform of independent senior legal talent and as an operating system for flexible, borderless legal work. So walk us through what that operating system provides to lawyers and clients.

Manuel Deó (11:55)

Sure. I sometimes regret using the word “platform,” although people sometimes refer to us like that, because people picture a database of lawyers with better typography, basically. We’re not putting lawyers on a website and calling it innovation, right? It’s a totally different thing.

Traditionally, a law firm bundled several things together. The talent, the brand, the clients, the technology, compliance, operations, and professional community. When a great lawyer became independent, that lawyer did not lose their expertise, but they often lost access to that entire institutional bundle.

What Ambar is doing is rebuilding those capabilities as a shared infrastructure without asking the lawyer or boutique law firm to give up their independence, their economics, brand, or professional identity.

For independent lawyers and specialist boutiques, which are the types of lawyers who are part of Ambar, we think about an operating system through four Bs.

One is belonging. You are an independent lawyer, but you’re also part of something bigger. Independence is professionally liberating when people leave certain institutions, but it also becomes lonely. We provide a curated professional community, practice groups, city chapters, trusted peers, knowledge-sharing spaces, and opportunities to collaborate with other exceptional lawyers. That’s the first one, belonging.

The second one is business. We provide access to sophisticated referrals among the different lawyers, cross-selling opportunities, team formation, and the infrastructure required to transform expertise into a sustainable legal business. Many outstanding lawyers do not want to spend half their lives becoming salespeople. That’s typically something lawyers don’t like to do, or chasing invoices, or learning digital marketing, or whatever. They want to practice law at the highest level. We help create those conditions around them to make that possible. So that’s the second B, business.

The third one is backbone. That’s the operational machinery around legal work. The intake, the proposals for the client, building the proposal with your scope, your fees, the contractual engagement letter, the specific conflict processes, KYC, anti-money laundering, insurance, invoicing, collections, all the red tape around sophisticated legal service delivery. We take care of that. The lawyer remains independent but no longer has to rebuild an entire institution alone with all the compliance and KYC around that. So that’s the third part, backbone.

The fourth one is the badge. That’s certification of your expertise, reputation, positioning, and institutional credibility, which is important in our sector and, obviously, in any professional service. We give independent experts and boutiques the ability to compete for sophisticated legal work that was historically reserved for larger organizations while preserving their own identity.

We are also building that with AI. AI is becoming more important in any profession, but obviously in legal. There are many players, frontier models building legal verticals, and AI legal-specific companies building legal tools. We have agreements with certain legal AI tools to deliver that legal AI expertise and those tools for our lawyers.

Greg Lambert (16:08)

Let me point at that last, almost like a fifth level here on the tech stack. How do you set up the attorneys on the technology stack so they have access to the tools, but then they also understand, say, how to use the tools, the interoperability between them, and all of that? How do you look at your tech stack?

Marlene Gebauer (16:28)

Responsibly.

Manuel Deó (16:34)

Right. I think we’re moving into a moment in which knowledge will be a commodity. That’s something hard for a lawyer because we’ve been, and I still consider myself a lawyer, tested for our knowledge of specific verticals, specific cases, specific contracts. But that’s a commodity.

I think the key here is that lawyers need to move from that sort of mantra where knowledge was key to a mantra where judgment will be the key driver of sophisticated legal services moving forward.

That’s our focus. Giving lawyers the ability to control, train, and use these tools, have full knowledge of the different AI tools available, and know that certain tools are available and helpful for specific needs while others are better for other needs. And second, providing judgment to that sort of legal service delivery.

The technology we’re building with other partners is allowing that judgment to develop memory, speed, and consistency, which are likely the three things clients are asking for.

Memory in terms of, “I want all my proprietary data, all my proprietary information to be stored. I want the AI to learn from my prior experience and from my knowledge.” Speed in terms of efficiency and time of delivery. And consistency. “I want to control that all my answers, all my legal advice, are consistent with my prior advice.”

That’s what we’re building. We’re building it with training, obviously, with different dedicated people in the different AI companies we’re partnering with. It’s not a one-shot process. It’s long.

We have more than 300 lawyers in our community currently from different backgrounds, different seniorities, and different geographies. I think sophisticated lawyers are typically tech savvy, but obviously not all of them have the same appetite to learn and the same ability to be aware of what AI is and what AI delivers to them.

Marlene Gebauer (18:59)

It’s interesting. You distinguish between knowledge as a commodity versus judgment, where I think a lot of firms would say their internal knowledge of things is kind of their IP. So how do you distinguish the two?

Manuel Deó (19:17)

There’s been a great debate around that topic lately in Big Law, basically because many firms have realized that if you hire an AI frontier model or AI legal solution, and you upload certain documentation to that AI tool, you might lose control over that information and not even know where that knowledge will end up.

So we’re building what we call the expert twins, which is a specific layer of information and documentation for each and every lawyer who is part of our community.

That’s basically a controlled, evolving system built so that the specific expert has approved knowledge. Knowledge that they approve, which is valid, which is dedicated. Picture a contract, picture a court case, something they’ve been validly involved in and they know that knowledge is valid. It’s been certified by them, with playbooks and experience specifically uploaded by that lawyer, and building a second layer of information validated by that lawyer.

That helps the lawyer trust the information being used by the AI tool.

Greg Lambert (20:46)

I want to switch over and look at Ambar from your clients’ perspective. I know a lot of them are looking at adding internal headcount or engaging with a traditional law firm for their legal needs. So how do you approach your clients and kind of sell to them what your platform at Ambar provides instead of these traditional methods?

Manuel Deó (21:15)

Sure. We use a simple rule. Fixed demand deserves fixed cost. Variable demand deserves variable capacity. Rare expertise deserves access and not ownership.

What I mean by that is that a company should hire internally, as an FTE, as a full-time employee, when the work is permanent, predictable, and central to its institutional knowledge, even typically sensitive core strategic information or knowledge in that organization.

But it should engage a traditional law firm when it needs institutional firepower, a highly leveraged team, court representation, a regulated local structure, or the resources required for a bet-the-company matter. M&A, capital markets, all those major, important transactions in the life of a company.

But Ambar sits between those two options when the need is important but variable, when they require senior expertise quickly, or when they need someone operating close to the business without making a permanent addition to the organization.

So you might have a fractional lawyer working 10, 20, 30 hours a week. It might be a leave of absence of one of your colleagues, post-acquisition integration, an internal expansion, a specialist regulatory problem, an investigation, litigation support for a specific period of time, or additional capacity during a peak of work.

In 2025, we handled approximately 300 projects, 293 projects, for 140 clients. More than 70% of that work was fractional counsel work, which is an external lawyer dedicated to your team as a natural extension of your in-house team.

This is important because fractional work is not simply a short secondment. The lawyer in a fractional project becomes part of the client’s operating rhythm, but the capacity remains adjustable. The client gains seniority, continuity, and business context without automatically adding permanent fixed cost.

Sometimes the problem is urgent, but that doesn’t mean the organization chart should become permanent. Clients do not always need another lawyer. They need the right expertise at the precise moment when that expertise becomes valuable.

More especially now, we’re in an uncertain moment where people don’t know, from a macroeconomic perspective, what’s going to happen. In many verticals and industries, there’s a lot of uncertainty. Of course, AI is bringing even more uncertainty in the legal sector.

Many clients, many general counsel, look at the fractional model as a quick, efficient, and flexible manner of accessing capacity without adding permanent headcount.

Marlene Gebauer (24:46)

So operating across jurisdictions with hundreds of independent lawyers, I picture, creates difficult questions around conflicts, privilege, professional ethics, data security, and AI use.

Greg Lambert (24:55)

A little bit.

Marlene Gebauer (24:59)

So how are all those responsibilities allocated and managed, and who’s ultimately accountable to the client?

Manuel Deó (25:11)

Sure. Independence must never mean ambiguity. In fact, a distributed model like ours only works when responsibility is more explicit, not less.

The allocation is clear. The lawyer remains professionally responsible for legal judgment, conflict checks, privilege, and all the ethical obligations applicable in the relevant jurisdiction in which they are barred or qualified.

Ambar is accountable for the operating environment we control. The certification of that lawyer, the selection, the contracting, the insurance we provide to our clients, the cybersecurity standards, governance, process, compliance, anti-money laundering, monitoring, and escalation.

In particular, one of the things you’ve mentioned is conflicts, conflict checks, which are a critical part of our business. Those are assessed in relation to the individual lawyer and the particular engagement.

We do not create one central environment in which independent lawyers access confidential information belonging to the clients of other lawyers in the community. We would never do that.

Each professional, each lawyer, retains responsibility for their own practice and professional obligations, supported by our engagement processes and infrastructure.

Privilege is addressed through the structure of the engagement and the rules of the relevant jurisdiction applicable to that engagement. It’s never something we simply assume because technology is involved.

AI use is also governed by clear principles of client requirements, approved tools, defined data boundaries, confidentiality, and human review. As you might expect, we have B2B enterprise clients who are sophisticated. They all have AI governance and data privacy policies which we need to comply with and be audited against.

Where local rules require work to be delivered through a regulated law firm or a particular professional structure, the engagement must reflect that.

So that’s what we do. We’re careful to do that and do not use the independence of those lawyers as a liability escape hatch.

We think distributed delivery should distribute capability, not responsibility. The client should never have to guess who owns the risk. We’re clear. The engagement letter covers that.

We have an engagement letter where the client, the lawyer, and Ambar are parties to the same relationship because we believe in transparency and each party knowing who owns what and who’s responsible for each part of the legal delivery.

Greg Lambert (28:21)

Well, I think this ties in nicely with the next question. I want to talk to you about Ambar’s community, its culture, and some of the interesting things I’ve seen.

As I was driving into work this morning, I was listening to something and someone mentioned culture at an organization. They defined it in a way that makes sense, but I hadn’t heard it quite this way before. Culture is a set of rules and guidelines on how a person should perform and act in the environment in order to succeed within this community.

When you have a distributed network of independent lawyers, let me ask you up front, how do you prevent this independent contractor from feeling like a transactional database of freelance attorneys? Let me start with that.

Manuel Deó (29:30)

Sure. It’s a long process. One is the right selection of the lawyers, so the vetting of that lawyer, being super demanding on the type of lawyer who is part of the community. We’re picky on that.

I hate to say that, but 95% of the lawyers who request being admitted to Ambar are not able to do it. So we have less than a 4% admission rate right now.

We require those lawyers to come from, or to have expertise in, a closed-ended list of law firms or in-house departments, which gives us certainty on the type of lawyer, the type of culture, and the work ethic around that lawyer.

We take that certification and admission process seriously. We’re part of it through the vetting of that lawyer with third-party references, but also with an in-person meeting with my co-founder, Rosa, who sits with each and every lawyer who’s part of Ambar and wants to be part of it. So we’re careful with that.

That’s one thing. Second, we have a clear code of practice. Obviously, when you’re talking about sophisticated lawyers, these lawyers come from specific backgrounds. They are subject to the ethical standards of their jurisdictions. This is a small market, and there’s tremendous ownership of each person’s reputation and brand.

We’ve never had any issue with that. We’re careful with it. We always have to be listening to our clients regarding the feedback we get on our lawyers in connection with our projects.

We’ve been disruptive there, too, because we’ve been proactively asking for feedback, which is something that’s not common in this sector, and getting the client to verbalize the feedback they have on their lawyers, put that on the record, and make that transparent to the lawyer.

So we’re careful about that. Obviously, it’s a distributed model. Ambar is accountable for the system and the lawyer is accountable for the legal judgment and the legal delivery. That’s clear. I think that’s something that’s also been clear for our clients.

Greg Lambert (32:24)

There’s one more thing you do in community that’s interesting to me, and that’s your communications with your clients.

You’ve put out a newsletter that I think you call Dr. No, and it’s read by thousands of general counsel. But instead of talking about case law updates and changes in regulation, you write about leadership, legal operations, and strategy.

How do your clients respond to a different approach like this?

Manuel Deó (33:10)

Sure. Dr. No, and by the way, the name comes from how certain in-house lawyers are described in their companies because they’re always saying no.

What we try to say there is that we need to put our clients at the center of the conversation, which is something that’s not common in the legal sector.

We’ve always been talking about the rainmakers in this law firm. We talked about Chambers before. Chambers has a lot of rankings, but they’ve always concentrated on law firms and not on in-house lawyers and general counsel.

What we’re trying to suggest here is that we should change the conversation, put the general counsel in the middle of the conversation, and offer different tools those lawyers in those companies need.

I’m not going to teach law to a general counsel or a senior lawyer in an important company. I would never intend to do that. But we talk about things they need to become more influential in their companies, to become more central.

Think about HR managers 20 years ago, who became influential people in their companies through many different changes. One of them is a change of name, which changed to Chief People Officer or Chief Talent Officer, and they became senior and influential people on the board.

Or CFOs, who went from being financial analysts or controllers to becoming influential people in their companies.

What we’re trying to provide are those tools for general counsel to become as influential as a Chief Legal Officer, someone who’s talking about metrics, return on investment, how much savings the department provides to the company, and treating a legal department as another separate business unit of a company.

Greg Lambert (35:25)

All right, well, we are at the end. We want to have you look back before we ask you our crystal ball question. So Manuel, what is something that’s different today, or wasn’t necessarily true a year ago?

Manuel Deó (35:48)

I don’t want to talk about AI because it would be too obvious. Obviously, every week we have a new development in AI.

Marlene Gebauer (35:59)

So boring, AI.

Manuel Deó (36:01)

Right. So it’s something I would avoid mentioning. It’s a difficult question, but I think we’ve run through a process where general counsel don’t want to be seen as a cost center in their companies and want to be seen as something different.

That’s changing the conversation, and it’s changing the type of solutions we as legal suppliers need to provide our clients.

Because it’s not a matter of an hourly rate or, “I’m the best expert in this specific industry.” It’s a matter of how much value did I provide to my client? Because I need to help that client explain to their CEO, their CFO, the value they’re bringing.

That’s something we’re starting to see now that we didn’t see a year, a year and a half ago. How important value contribution is, and being a true investment as a legal department in a company, is changing the mindset of many general counsel.

Marlene Gebauer (37:11)

Well, if that question was hard, this one might be a little harder.

Greg Lambert (37:14)

Yeah.

Marlene Gebauer (37:15)

This time we’re going to ask you to look into the future.

Manuel Deó (37:17)

Wow.

Marlene Gebauer (37:18)

Our crystal ball question is, what’s the single biggest change you see coming for how firms operate and interact with their clients?

Manuel Deó (37:29)

That’s also a hard one, yeah. I think the single biggest change will be that traditional firms will stop being the only meaningful unit of legal delivery.

Law firms will be important, but they will become more porous. The future legal team will combine small strategic in-house core teams. They will engage traditional firms when they need them, but they will also engage independent senior specialists, specialist boutiques, managed services, and AI agents to deliver legal solutions appropriate to the specifics of that legal need.

Those capabilities will be assembled by each legal department in a dynamic manner around each problem.

I think clients will move from asking, “Which firm should we hire?” to, “What is the best configuration of people, technology, knowledge, and risk for this particular matter?”

The traditional pyramid was designed for a world in which legal knowledge was scarce and coordination was expensive. That will be totally different.

AI makes information, as I said before, increasingly abundant and truly a commodity, in my opinion. Technology makes coordination much cheaper. I think that will change the geometry of the industry.

I think we’ll be seeing legal departments more as trusted orchestrators of capabilities. That’s how I see the future going forward. AI, boring matter, but I know it’s something we always...

Greg Lambert (39:26)

Well, I think you handled both of those questions well.

Marlene Gebauer (39:27)

Yeah, I think you did a great job.

Manuel Deó (39:28)

Yeah.

Greg Lambert (39:30)

Well, Manuel Deó, thank you so much for taking time out of your day to talk to us. This has been great.

Marlene Gebauer (39:37)

Thank you.

Manuel Deó (39:37)

Thank you. Thanks for having me.

Marlene Gebauer (39:40)

And thanks to all of you for listening. And don’t forget to like and subscribe on Substack, YouTube, and LinkedIn.

Greg Lambert (39:48)

And Manuel, where’s the best place for people to follow your content and to learn more about Ambar?

Manuel Deó (39:54)

Well, the best place to follow me is LinkedIn. I write regularly about the business of law, leadership, innovation, entrepreneurship, and AI.

I also publish a daily newsletter called Dr. No for legal leaders who want to be more innovative, about mental models, about cool stories. Less than a one-minute read each and every day.

Many clients tell me they resend it and forward it to their teams to send specific messages.

And that’s pretty much it. You can find Dr. No on our webpage, ambarpartners.com, under Dr. No.

Marlene Gebauer (40:38)

And as always, the music here is from Jerry David DeCicca. Thank you, Jerry, and bye, everybody.

For millions of people, an everyday legal dispute never justifies the cost of a lawyer or a private mediator, no matter how much the outcome matters to them. Judge Victoria Wood saw that problem over and over during her years on the Napa County Superior Court bench. This week, Wood joins Judicaid Chief Strategy Officer Valerie Clemen to explain how those years led to Judicaid, an AI-assisted mediation platform built to help people resolve everyday disputes before time, expense, and emotion push them deeper into litigation.

Wood traces Judicaid’s origins to two problems she kept running into as a judge and mediator. Traditional settlement conferences arrive late in a case, after the parties have spent real money and dug into their positions. Then there are the lower-value landlord-tenant, contractor, neighbor, probate, and small-claims disputes where professional mediation rarely pencils out at all. Wood puts a number on the access problem. The State Bar of California’s 2024 Justice Gap Study, released in 2025, found that Californians received no legal help, or inadequate help, for 85 percent of their civil legal problems. Judicaid aims at a slice of that gap by giving people an earlier and cheaper chance to communicate and negotiate.

Clemen walks Greg Lambert and Marlene Gebauer through Judicaid’s “shuttle-style” mediation process. Each participant talks privately with an AI mediator named Jude, so the two sides never have to speak to each other directly. Jude gathers each side’s account of the dispute, identifies priorities and possible settlement terms, and moves between the participants while filtering out insults, anger, and inflammatory language. If the parties find common ground, the platform prepares a proposed settlement for review and electronic signature. Wood is careful about one boundary throughout the conversation. Jude is a facilitative mediator, and it stays away from evaluating legal rights. It will not decide who is legally correct, predict who will win, or give legal advice.

The conversation then turns to Judicaid’s pilot with Napa County Superior Court, and the role courts could play in expanding AI-assisted dispute resolution. In the court model, a court subscribes to the service and hands litigants’ access through a QR code, with no integration into the court’s technology systems. The pilot has already surfaced a behavioral lesson. Offering mediation as an optional service does not mean parties will use it, and Wood and Clemen see more potential where courts actively encourage or require litigants to attempt dispute resolution before proceeding. Language is another piece of the story, since Judicaid lets participants who speak different languages work through the same mediation without arranging multiple interpreters.

Wood and Clemen also see mediation as just the starting point. Wood uses the phrase “intelligent dispute resolution” for a broader category of AI-assisted tools covering mediator proposals, parent coordination, and other structured approaches to conflict. The bigger ambition is a change in habits, where people reach for structured communication and settlement before a disagreement hardens into a lawsuit. Clemen boils that aspiration down to three words she hopes become part of the vocabulary of everyday disputes: “Just Judicate it.”

Listen on mobile platforms:  ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Apple Podcasts⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ |  ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Spotify⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ | ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠YouTube⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ | ⁠Substack⁠

[Special Thanks to ⁠⁠Legal Technology Hub⁠⁠ for their sponsoring this episode.]

Email: geekinreviewpodcast@gmail.com

Music: Jerry David DeCicca

LINKS

• Judicaid
  https://judicaid.com/
• State Bar of California, California Justice Gap Study
  https://www.calbar.ca.gov/about-us/data-reports/view-all/california-justice-gap-study
• 2024 California Justice Gap Study
  https://www.calbar.ca.gov/report/2024-justice-gap-study-report
• Napa County Superior Court, Small Claims
  https://www.napa.courts.ca.gov/divisions/small-claims

Transcript

Marlene Gebauer (00:00)
Hi, I’m Marlene Gebauer from The Geek in Review, and this week we’re talking with Judge Victoria Wood and Chief Strategy Officer Valerie Clemen from Judicaid. But first up, here’s a word from our great sponsor and partners at Legal Technology Hub.

Marlene Gebauer (00:13)
I have Sam Moore here from Legal Technology Hub. He’s going to tell us about AI governance from the advisory side.

Sam Moore (00:19)
Thank you. Good to see you again. Well, in the advisory team at Legal Tech Hub, we’ve been having a lot of conversations with clients about AI governance. Now, most law firms and departments have at least an AI policy of some kind, and there are some broad standards starting to emerge. A trend we’re seeing at the moment is a shift away from policies written in 2023 and 2024, which largely stated what you couldn’t do, and toward a more informative approach, which

Marlene Gebauer (00:37)
Okay.

Sam Moore (00:47)
defines different risk categories and

Marlene Gebauer (00:48)
You...

Sam Moore (00:49)
makes distinctions between routine, low-risk use of AI and other more substantial use cases. I think that’s appropriate for where we are, and it’s also what clients are coming to expect from their advisors. But one of the bigger challenges we’re seeing right now is how you express an AI use policy in such a way that any member of your legal team could explain it to a client. I think that’s still a big challenge for the industry.

Marlene Gebauer (00:55)
So, okay.

Sam Moore (01:17)
And I don’t think clients are all that happy if their attorney says, “I’ll have to get back to you about that,” because the attorney is probably using AI day-to-day. So they should be able to explain the policy themselves. How else do you know they’re complying with it? We’re doing some interesting work at the moment helping law firms and law departments take those first-version AI use policies that are very much “thou shalt not”

Marlene Gebauer (01:18)
So...

Sam Moore (01:43)
and turn them into more readable, appropriate governance positions that inform the conversation with clients. If anyone wants to find out more about our advisory services, they can find me or Cheryl Wilson Griffin on LinkedIn, or they can visit https://legaltechnologyhub.com.

Marlene Gebauer (02:01)
Thank you, Sam. I mean, it is really important to have AI policy in plain language so people understand it.

Sam Moore (02:07)
Absolutely.

Marlene Gebauer (02:15)
Welcome to The Geek in Review, the podcast focused on innovative and creative ideas in the legal industry. I’m Marlene Gebauer.

Greg Lambert (02:22)
And I’m Greg Lambert.

Marlene Gebauer (02:24)
Today we’re exploring whether artificial intelligence can help address one of the justice system’s most persistent problems, resolving everyday disputes without months of delay, escalating costs, or procedures that ordinary people struggle to navigate.

Greg Lambert (02:39)
Yeah, and to help us do that, our guests today are the Honorable Victoria Wood, a retired judge from Napa County Superior Court and founder and CEO of Judicaid. Joining Judge Wood is Valerie Clemen, Judicaid’s Chief Strategy Officer. Judicaid is an online, AI-assisted mediation platform designed to help people resolve disputes privately, affordably, and without first filing a case.

Marlene Gebauer (03:08)
Judge Wood and Valerie, welcome to The Geek in Review.

Valerie Clemen (03:12)
Thank you for having us.

Vickie Wood (03:12)
Thank you. Happy to be here.

Marlene Gebauer (03:16)
Judge Wood, Judicaid grew out of what you witnessed during your years on the bench in Napa County. What recurring problems convinced you that the traditional court and mediation systems were not adequately serving people with everyday disputes, and that technology could provide a better path?

Vickie Wood (03:34)
There were a few things. One of them, I’d say, is that I handled a lot of settlement conferences while I was sitting on the trial court bench, but usually those come very near trial, later in the case, when people are very entrenched.

Marlene Gebauer (03:50)
When the judge says, “You go in the other room and figure it out.”

Vickie Wood (03:54)
Yeah, yeah.

And they’ve already become very angered in their positions. They’ve already invested so much emotionally and financially, usually through attorneys’ fees, and it’s hard to get those settled. Sometimes that’s what it takes because people understand the dynamics and the finances, and they’re finally at their wits’ end and they’ll settle. But it comes at a cost. There may be families that could have had relationships saved if they could have talked earlier on and understood each other’s positions better.

So that’s one thing, the opportunity for earlier mediation or communication facilitation. For example, I handle a fair amount of probate disputes, and some people say probate disputes are essentially family law after someone dies. They’re highly emotional for lots of reasons that go back many years. But what very often happens is it seems like there’s a lot of money at stake, but by the time they go through the process and get to mediation, sometimes the entire amount is spent on fighting. And it’s really sad. People never thought that would be the case.

I often think to myself, boy, if I only could have been with these people earlier in the process and let them know what would likely happen if they continued down the litigation path. So that’s one of the things that inspired it. The other thing is that there is this broad swath of disputes that are lower value and really don’t justify the expense of resources for attorney representation or human mediators. But those disputes still mean a lot to people, even though they’re low-dollar-value. In fact, that’s the reason why I decided to go to law school, being in a very low-dollar-value dispute with a landlord who I felt had done us wrong, and I was gonna

Greg Lambert (05:56)
Ha ha.

Vickie Wood (05:57)
fight it. And we did sort of win, but boy, it took a lot of time.

But these cases mean a lot to people, and they don’t have the same access to alternative dispute resolution as higher-value cases. So that was the other problem I was looking to fix. In fact, there is a State Bar of California justice gap study that was recently released. It was a follow-up study from 2019, then redone in 2024, with those results coming out in 2025. They reflect, let me get this right, that Californians receive no legal help or inadequate legal help for 85% of their disputes.

And I thought that was astonishing.

Marlene Gebauer (06:59)
I wish I thought it was astonishing. It’s like, I...

Greg Lambert (06:59)
Yeah. Yeah. And...

Marlene Gebauer (07:03)
I think...

Greg Lambert (07:03)
I don’t think Californians are unique in this either.

Marlene Gebauer (07:06)
No, no.

Vickie Wood (07:07)
Yeah. So this was a California study, obviously, but that’s 85% of disputes where people don’t have help, don’t have adequate help, or they throw their arms up and say, “The justice system can’t help me.”

After the 2019 study, the California Judicial Council really made efforts to try to close this justice gap. The problem was that the results from 2024 showed things got worse, not better, despite their efforts. So that was striking to me as well.

When I left the bench and started my mediation practice and started studying and using AI, it came to me that there is probably a portion of these cases that could benefit from what AI and LLM technology can do. That’s how Judicaid was born. I came up with the idea and the name, and then I happened to have a family member who is brilliant in the tech world. I mentioned it to him and he said, “I can build that for you.” And he said, “That will be easy.”

So, more than a year later...

Greg Lambert (08:27)
Yeah, I imagine in California...

Marlene Gebauer (08:28)
Words you like...

Greg Lambert (08:29)
Yeah. Well, it’s nice...

Marlene Gebauer (08:29)
to hear when you’re building tech, “It will be easy.”

Vickie Wood (08:32)
Yeah, yeah. “That’d be easy.”

Greg Lambert (08:33)
to be in California where you can’t swing a cat without hitting an engineer.

Vickie Wood (08:39)
Yeah, needless to say, it was not easy. It proved to be much more challenging, but he and the other part of the tech team did it. We’re really proud of the tech product, the rest of our product, and the platform, and excited to be getting it out in the world. So thank you for the opportunity to be here today.

Greg Lambert (09:00)
So, Valerie, I want to jump into what Judicaid does and how it’s set up. But let’s start off with the user experience. Let’s take a typical user of the system, maybe two people having a landlord-tenant, contractor, or small-claims dispute.

What happens from the moment one person opens a matter on Judicaid until the parties either reach a settlement or determine that the mediation won’t resolve it? Can you walk us through the process of Judicaid?

Valerie Clemen (09:37)
Sure. The product just went live, I want to say, like six weeks ago. There are actually two aspects to it. There’s the consumer-facing version, where people can go on, pay $34.99, and start a mediation. But we also have the court-subscription version. Napa Superior Court currently has a pilot in its small-claims cases, so it’s a little different depending on which one we’re talking about.

Essentially, it’s a website, a standalone website. Parties can go there, create their own account, and put in some basic information about who the other side is and what the dispute is about. They never speak directly with the party they’re having a dispute with. An email goes to the other side and says, “We’re sorry you’re in a dispute. The other side would like to try to work this out with you. Please sign up as well.”

Hopefully they sign up if they’re willing to talk. Once they sign up, there’s literally an AI mediator that first goes to the initial party and goes back and forth asking questions like: When did this happen? What occurred? What is owed? Usually it’s over money or property. What attempts have there been to resolve it? What do you want today to resolve it? Once it has this information, it says, “Sit tight,” and then it goes to the other side.

When it says “sit tight,” people can be doing this on their phone, laptop, or tablet, and they don’t have to be online at the same time. It basically sends a message to each side when it’s their turn to speak to the AI mediator. It’s essentially the LLM, the AI engine, asking the questions and doing this.

So it goes to the other side and asks the same kinds of questions. What’s your version of the dispute? How do you feel about it? What’s your bottom line? What do you want? Once it has that, it starts going back and forth and tries to ask questions, get more information, and bring both sides to an agreement, hopefully in the end. It’ll keep going back and forth as long as it takes.

We’ve done a lot of pressure testing. Vickie and I are both attorneys, so we were like, what would a litigant say if you’re really upset? They might tell the AI mediator to F off and say, “You’re an idiot,” or whatever. So we did all of these things, and it’s great. It’ll hear these things and say, “I understand you’re upset, but what is your bottom line?” It’ll never go to the other side and say that this side insulted you. It’s very good at filtering

Marlene Gebauer (12:16)
Yeah.

Valerie Clemen (12:17)
and focusing on the problem and how to resolve it. Once it senses there’s an agreement, maybe there’s a payment plan worked out, it’ll prepare a draft agreement and send it to both sides and say, “Could you live with this? Is this what you want? Is this the agreement you want?” If both sides say yes, it creates an agreement, sends it to both sides through DocuSign, and you have a signed settlement agreement.

Greg Lambert (12:43)
Okay. Well, let me jump in on one question. What incentive does the counterparty have? I can see an incentive where I want to pay the $35 to get the subscription, but what’s the incentive on the other side to join this mediation?

Valerie Clemen (13:08)
Usually there’s a threat of a lawsuit. Like, “I’m gonna take you to small claims. I’m gonna call the Contractors State License Board.” If it’s a home-improvement project, there’s some threat because someone’s upset and wants something. So either you resolve it with that person or they’re gonna take the next step, which could have professional consequences. You might have to deal with litigation. You’re gonna be sued. In the world of Google, if you get sued by someone, somebody Googles your name and that shows up forever.

So the ability to resolve something before it turns into something worse would hopefully be motivation. But, you know, humanity...

Vickie Wood (13:46)
Yeah, but Greg, you ask a really good question because that is a sticking point. The platform generates an automatic email invite to the other side, and it says this person has either filed a lawsuit or is considering filing a lawsuit, but they would like to first try to resolve this peacefully.

You don’t lose any of your rights if you’re not able to resolve it. I personally would be very motivated to try that, knowing what it’s about, but what we’re finding is that a lot of people don’t even know exactly what mediation is, and they don’t know what they’re getting into. We try to highlight that this is completely confidential. The court’s never gonna know anything that is said. You preserve all your rights.

But that is a tricky part of this, and we’re hoping that if it gets out there and becomes more normalized, that’ll take away some of that reluctance. On the other side of the coin, I think people in general would really like to avoid going to court, seeing this person face to face, and talking to a judge. These are scary things. People are getting less used to talking to people face to face, so we’re hoping it’s going to play into the natural progression of how people would rather address conflict.

Marlene Gebauer (15:18)
So I’m curious about how the AI is trained. I’m assuming that it trains on contract law, landlord-tenant law, things that would come up as part of these disputes. I’m also curious, we talk about personas here sometimes in terms of different types of training and things like that. I’m wondering if judge personas were part of the training.

And then how do you keep the system from crossing the line into giving legal advice? I heard a little bit, if we sense there’s an agreement, we’ll start with the decision, but how do you avoid legal advice or saying, okay, right, wrong?

Vickie Wood (16:03)
Mm-hmm. Great questions. First off, no, it’s not trained in the law.

It doesn’t give any legal advice whatsoever. It is essentially a facilitator of the communications. As a mediator, most mediators know there’s a spectrum of how you approach mediation. You can be on one end, mostly facilitative, where you’re exploring people’s values and priorities in trying to get the case resolved, and you’re facilitating communication about how they feel and their thoughts on the case.

On the other end of the spectrum is a mediator who becomes quite evaluative and says to someone, “You’re gonna lose this case,” or, you know, “Here’s...”

Marlene Gebauer (16:48)
You know, “You might want to think about this.”

Greg Lambert (16:50)
Mm-hmm.

Vickie Wood (16:52)
“This is where I’m really concerned. These are your risks.” Generally, at least in how I mediate, and I think it’s one of the better ways to approach mediation, you start out facilitative and then, as the day goes on, you can get a little more evaluative and try to push things toward that end.

Judicaid stays facilitative. So there’s no legal advice. It’s not trying to tell someone, “You’re gonna lose this case,” or, “They’re right on this point and you’re wrong,” or vice versa. It will remind them of the risk and hassle, the time and expense of going to court, and the risk that they may not win their case, encouraging them to continue to think about compromise and keep engaging so they inch their way toward a palatable compromise where eventually they go, “Okay, I can say yes to this and be done and not have to deal with court.”

So that’s the idea, and that’s why we’re able to offer it nationally, because it doesn’t deal with laws that would be individualized to each state.

The training aspect is really, there are two main tech pieces. There’s the application that’s coded, which is our AI engine for the shuttle-style back and forth. That took a really long time to create. Then there is the mediator disposition architecture, and that’s where I had a lot of input as the subject-matter expert. It’s like a 38-page prompt that...

Marlene Gebauer (18:42)
Ha ha ha.

Vickie Wood (18:42)
That...

Greg Lambert (18:42)
How?

Vickie Wood (18:43)
guides Jude, our mediator’s name is Jude, by the way. It tells Jude...

Marlene Gebauer (18:48)
Do you say, “Hey Jude,” when you talk to it?

Vickie Wood (18:51)
Jude. We actually...

Marlene Gebauer (18:53)
I’ll leave now.

Vickie Wood (18:54)
never intended to have a name, and it just sort of came about. Since it’s Judicaid, and Jude’s sort of a gender-neutral name, we thought, why not? We’ll go for it.

But it’s the disposition, sort of prompting architecture, that tells it how to respond and how to continue to inch people toward a compromise. There are lots of guardrails so it’s never giving legal advice. It can suggest, if someone says, “Do you have any ideas on how we might be able to resolve this?” it can say, “Well, one thing you might consider is this,” or, “You could take payments over time,” or, “Maybe take a smaller payment up front and then the balance over time.” It can suggest things like that, but it’s very generalized in that way.

That’s a big part of why it was never intended to replace attorneys or human mediators. As a mediator, I know that most complex cases that go to human mediation are far too complex, both legally and emotionally, for an LLM to manage. So this is not trying to replace that. This is focused on filling the gap for those lower-value cases that are generally fairly simple.

There is emotion involved, and that’s something an LLM is very good at managing in communication. It’ll calm the person down, and they can get as upset as they want. It can be empathetic and say, “I hear you’re feeling really frustrated about that, and that’s understandable. But why don’t you let me go talk to the other person?” It’ll go to the other side, and it’s never gonna convey the anger or the bad words being said in the other room. It’s gonna say, “They’re feeling concerned about this.” So it paraphrases things in a way that makes them far more palatable to hear when you’re the adversary.

That’s a big part of what I do as a human mediator. When somebody is cussing and swearing and saying horrible things about the other side, I go in the other room and say, “Yeah, well, they’re starting to come around.” So...

Greg Lambert (21:17)
Yeah, I was gonna say, do you have it default back to those instructions so it doesn’t get caught up in what each side is feeding it, but rather goes back to its baseline to make sure it stays on the course it needs to stay on?

Vickie Wood (21:40)
Yeah. Like I said, the prompting architecture is vast, and it’s very carefully thought out in phases to guide things through. But it has priorities around guardrails, and if there’s any sort of safety issue, somebody says, “This is terrible. I’m gonna kill myself,” we make sure it’s gonna say, “If you’re struggling with that, then you need to consult resources in your locality. This may not be the place to do this.”

And there’s a button to say, “Stop mediating,” and terminate the mediation at any point in time. So, yeah, a lot of thought went into that...

Marlene Gebauer (22:25)
I have one. I have one.

Vickie Wood (22:27)
Aspect.

Marlene Gebauer (22:29)
Does it learn? I mean, from session to session or are those isolated?

Vickie Wood (22:34)
They’re isolated. We have made the decision at this point that we’re not training the AI model directly in the way that an open-source LLM would be constantly training itself.

Yeah, it’s truly intended to be confidential, just like mediation is confidential. You’re not supposed to go tell a court, “Well, in mediation she said this or that,” or you’re not allowed to, at least in California.

Greg Lambert (23:03)
And that’s when you hold up your hand saying, “I don’t want to hear that. I don’t...”

Vickie Wood (23:06)
Yeah. Yep. Exactly.

Marlene Gebauer (23:08)
No, no.

Vickie Wood (23:09)
Exactly. Privileged. So...

Greg Lambert (23:12)
Well, we talked about the user side, so let’s go over and talk about the court side. Courts appear to be a really important path for Judicaid going forward. Valerie, I’ll start with you, and Judge Wood, if you want to jump in, please do. What would a court-facing model look like in practice, and where do you see it fitting into the life cycle of these disputes?

Is it before the case is filed? Is it after, but before a hearing? Is there someplace else? Where do you see the path of leveraging either the threat of the court or the actions of the court to help Judicaid?

Valerie Clemen (23:55)
Yeah. Initially, it would be good to have it in self-help centers, because a lot of times, especially for small claims, landlord-tenant, or those types of claims, those parties are unrepresented. You just can’t afford to pay an attorney’s hourly rate for a case that’s honestly below $20,000. It doesn’t make any sense to pay an attorney, so they’re trying to figure out the system themselves.

Knowing they have this option to resolve their dispute without going through the paperwork, getting it served, and filing is helpful. So having that information in self-help centers and in the courts pre-filing is helpful. Once a case is filed, it would be helpful for courts to provide that option and, honestly, not make it voluntary. Make it mandatory to give it a try, to talk to the other side using some type of dispute resolution or mediation, but also provide Judicaid as a free resource for litigants to use and try to resolve the dispute.

The information would come at the time of filing. Like, here’s your complaint, you have to serve the other side, and also serve the other side with this dispute-resolution information to see if you can resolve the case before your hearing date comes up. In some courts, when your hearing date is set out six months or a year, being able to resolve it and put the thing behind you before that happens, or before you spend money having the sheriff personally serve it or hiring a process server, is helpful. There are so many costs and so much time associated with litigation.

The hope is that letting litigants know there’s this option would be helpful. Of course, a lot of unrepresented litigants are doing this for the first time. They’ve never filed a small-claims suit. They’ve never been in a legal dispute. A lot of them probably don’t even know what mediation is. I’d say at least 10% get the flyer and are like, “Meditation? I know I’m angry, but I don’t need that.”

Marlene Gebauer (25:48)
Yeah.

Valerie Clemen (25:49)
Like, just not having any idea.

Vickie Wood (25:51)
It should go hand in hand, though.

Greg Lambert (25:51)
I need it, but not in this case.

Valerie Clemen (25:54)
Yeah. Because honestly...

Greg Lambert (25:57)
So are you educating law librarians and self-help centers on this, so they have it as a tool they can point people to as an alternative, or...?

Vickie Wood (26:11)
We’re trying to. We’re trying to. We just...

Marlene Gebauer (26:15)
That’s why they’re on the podcast, Greg.

Vickie Wood (26:18)
But yeah, in Napa County, where we’re running the pilot, we have definitely put it out there to all of the people who might be interested. We’ve got flyers in the law library and in the regular county library. While it’s being offered for free, government agencies have a line where they don’t want to be advertising something that’s for profit.

But while they’re piloting it, or if they were, and actually, let me jump back to one of your questions, Greg, about what this would look like through the court. The platform itself will be identical. The courts would pay for a monthly subscription, and then they would be given a QR code that they could provide to litigants upon filing. They could even put it on their website, in the small-claims area, for people who are thinking of filing a claim to try it before they file.

Those people would enter the site through a landing page for that court with that QR code. There’s a little gateway where they have to say how they got there. If they didn’t get there through that court, we’re going to redirect them to the paid portion. But if they got there through the court through the QR code appropriately, then they bypass the payment piece and jump into the platform exactly the way everybody else does.

So, yeah, we’re trying to get this out there. We’re on social media. We just got back from a national court conference in Delaware. We were having conversations with courts across the U.S., getting them aware. I think the trick in trying to get courts on board with this is that most of the technology out there needs to be integrated into the courts’ networks and processes.

Their first thought, I think, is, “Here’s yet another AI-based app or process that we’re gonna have to implement and then be tied to this company for however long, and tied to their schedule of upgrades and new features and their pricing changes and all of that.” We’re completely different from that. The only thing the court has to do is give that QR code to its litigants or potential litigants, and any questions go through our support page. That’s it. That is it.

I think courts are having a hard time comprehending that it could be that easy. Hopefully we’ll get the message out. What was that, Marlene?

Marlene Gebauer (29:01)
This could be very helpful for them.

Greg Lambert (29:02)
Yeah.

Marlene Gebauer (29:05)
It could be very helpful for them in terms of the docket. So...

Vickie Wood (29:11)
Yeah, that’s the idea. And it’s so affordable that it’s sort of a win-win-win proposition, and it feels like a no-brainer to us. But still, it’s gonna take some time to get people to first be aware of it and then have a true understanding of how simple it is.

Valerie Clemen (29:29)
Yeah. I’ve had clients come to me, we’re in a small town, so I have clients who come to me with certain issues, and sometimes they come with things that are just so minor. They have a dispute with their neighbor and the money involved is so low, and I’m like, “Look, I can’t honestly have you pay me for this, but try this.” I’ve had a couple of clients go try Judicaid, and they were like, “It was easy to sign up and send the invite, but the other side didn’t respond.”

But it was nice. Sometimes clients come to you and they have things that are too minor, and you’re like, “Look, the budget for this doesn’t make sense.” Giving them something, some tool where they might resolve it on their own, was nice instead of saying, “I can’t take this. Good luck.”

Vickie Wood (30:12)
Yeah. So that is another sort of distribution channel, if you will. Exactly what Valerie just said, for attorneys to be able to give something of value to those clients they have to turn away. Say, “There’s another tool here.” We think it’s something attorneys would really like having in their back pocket for that situation. The problem is there is that immediate reaction of, “What is this taking? Is this going to take away our business?” Or, for...

Greg Lambert (30:45)
Yeah.

Vickie Wood (30:45)
mediators, “Is this going to take away our business?”

And I’m saying, no, it’s not that. LLMs are obviously changing the legal scene drastically, but at this level of sophistication, they can’t replace humans. You absolutely need human mediators. So hopefully, if there are attorneys out there listening, this is something for you to have to give to the people who you can’t otherwise help.

Valerie Clemen (31:16)
Yeah. It’s like...

Marlene Gebauer (31:16)
Yeah, I’m kind of looking at it like it’s an initial step because you’re absolutely right. I was very curious, a human mediator can sense when there’s unequal bargaining power or someone’s more legally sophisticated, and they can run their mediation with that in mind. But I’m curious how Judicaid does that as a technology tool.

Vickie Wood (31:43)
Well, in some ways it’s more neutral. Because it’s just going to be kind of...

Marlene Gebauer (31:47)
Yeah. It’s neutral. That’s it.

Vickie Wood (31:54)
Yeah, yeah. It’s not gonna be impressed by the person who is sort of a bully or sounds like they really know what they’re talking about. It’s not gonna be more persuaded by this side or that side. Again, it’s facilitating that conversation in a really neutral way.

And to your point about it being a starting point, yes, I do think cases will get resolved through this. But if they don’t, I also think there’s another value that goes beyond monetary. Let’s talk about this, it isn’t limited to small claims. It could be used for limited civil cases. It could be used for eviction cases. We’re gonna add a feature to try to get this into family law for parent-coordination kinds of things. I have unlimited features in my head, so it’s just a question of getting them to happen.

What I would see with small claims is, like I said in the beginning, people are passionate about these cases and they spend days’ worth of time preparing for them and putting everything into them. Then they get to court and there are 40 cases on the docket for trials. They’re gonna sit there for a long time, their case gets called, and they get maybe 10 minutes. And...

Greg Lambert (33:13)
Yeah.

Vickie Wood (33:14)
even if they win, but especially if they lose, they walk away not feeling good about the process and about the justice system, because many times this is these people’s only real interface with the courts. They walk away not feeling good about it, and that’s not good for any of us. That’s not good for society in general.

So if they start with a product like Judicaid, where they’re able to kind of be “heard,” and I’m putting that in air quotes for anyone who’s only on audio, to air their feelings about it and also hear the other side’s position and thoughts, and then have to think through the potential outcome that it may not go their way and decide they want to take that risk, well, now they’ve had much more participation in the ultimate outcome. I feel like that’s going to be an improvement for everybody across the board.

Greg Lambert (34:12)
Yeah. Well, I want to touch on your pilot with the Napa County Superior Court. Tell us about what you’re piloting. Beyond finding out whether the technology works, what are you hoping the courts will find out, and what are you hoping to find out from this pilot on your side?

Vickie Wood (34:35)
Yeah, so we’re new with the pilot. We’re kind of just starting to get through the period from where we started to where cases would be coming on to trial. We have had people engaging, but we’re not getting the mediations to happen yet.

So we’ve learned one really important thing, which is that people are gonna be reluctant to do this right now if it’s purely voluntary. That was a very valuable thing to learn. We knew there would be reluctance, but there’s significant reluctance.

One note, though, is that Napa County is a tiny county, so the number of filings is a very small sample to be working off of.

Greg Lambert (35:18)
I’m envisioning the mediation being over a glass of wine. That’s all I see.

Vickie Wood (35:22)
Of course, Greg, of course.

Marlene Gebauer (35:22)
Ha ha.

Vickie Wood (35:25)
If they’re not drinking wine, then the dispute is about wine somehow. But yeah...

Greg Lambert (35:30)
Yes.

Vickie Wood (35:32)
It’s a small county. But yeah, once we start getting engagement, and we’re connecting with some larger counties, one in particular is a large county that already mandates mediation for both small claims and unlawful detainers. They offer some human mediation options that are lower-cost than most human mediation, but still significantly more than Judicaid. So this would be another thing to offer people.

One of the greatest features about Judicaid is that it can be done in many different languages. You can have somebody who speaks only Spanish and someone who speaks only Russian mediating together. That would be very expensive to provide with a human mediator because you would probably have to have two interpreters for one mediator, and the cost skyrockets.

So, yeah, we’re hoping to get a bigger sample, especially one where it’s mandated, because we think if it’s mandated, a lot of people will opt for this because it’s so easy. You don’t have to schedule a set time. You can do it literally over your phone, dictating.

Greg Lambert (36:47)
And you mentioned that this isn’t just California, right? You’re looking at going...

Vickie Wood (36:51)
Correct.

Greg Lambert (36:54)
going into other states as well.

Vickie Wood (36:56)
Correct, correct. Yeah. And maybe, I mean, honestly, I don’t know why it couldn’t go outside the U.S. too, because again, it’s not tied to any particular laws. So I don’t know. I don’t want to think too big just yet, but...

Greg Lambert (37:09)
Well, then you have to deal with the EU and GDPR and all that, so, you know...

Valerie Clemen (37:14)
Yeah, yeah.

Vickie Wood (37:15)
Yeah. Well, I’ve been really bored lately, Greg. That would give me something to do with...

Greg Lambert (37:19)
Yeah.

Vickie Wood (37:19)
my time.

Marlene Gebauer (37:19)
Yeah.

So we kind of close the podcast by looking back and then looking forward. This is kind of our “what’s different now” question. If we had spoken with you a year ago, other than the fact that maybe Judicaid didn’t exist, what would you have told us is materially different today? What have users and court partners, or the experience of building the platform, taught you that changed your original assumptions?

Vickie Wood (37:56)
When I conceived of it, which was closer to two years ago, the way ChatGPT or any of these large language model platforms functioned was astonishing. I mean, it still is astonishing, but more people were seeing it for the first time or had not seen it yet. So the concept of what Judicaid does was almost like magic. Now people are aware that having a computer talk to you like it’s a human can be done.

What we have, that I don’t think is done anywhere else, is this shuttle-style approach where it’s truly exchanging the communications. But when my CTO told me, “This will be easy,” I thought we would be launching in like two months. Then all these months later, it’s not quite as novel as it was at the time.

At the same time, there’s also a fair amount of skepticism of AI and concern about AI taking jobs. There are people who might see that it’s an AI-facilitated platform and want nothing to do with it, without considering that this is something trying to provide a positive benefit and fill a gap. Those were unexpected things, but we’re rolling with it because I think it’s gonna require a little bit of awareness about what we’re doing.

Also, AI is becoming so pervasive that I think eventually people are...

Greg Lambert (39:41)
Yeah.

Vickie Wood (39:42)
just gonna get tired of being so against it, I guess. But yeah, I think the other thing is we focused on mediation because that’s the piece we wanted to nail down. Now that we have that core AI engine, we can branch off in so many different dispute-resolution ways, which gets me so excited.

We’re now calling it intelligent dispute resolution because it may not just be mediation. We might start having mediator’s proposals. We might have a parent coordinator to help with little custody disputes.

ODR, online dispute resolution, has been around for quite some time, but I think intelligent dispute resolution is new. So I’m gonna say you heard it here first.

Greg Lambert (40:29)
All right, all right. Well, that leads in perfectly to our crystal ball question. Valerie, I’ll let you start this off if you don’t mind. So...

Valerie Clemen (40:38)
Sure.

Greg Lambert (40:39)
Picking up on Judge Wood’s discussion there about intelligent dispute resolution, in three to five years, where do you see things when it comes to AI-assisted dispute resolution? What’s it gonna look like, and what all do you think it will cover?

Valerie Clemen (40:58)
I mean, our hope is that it would be offered in courts nationwide, because this underserved population doesn’t have access to dispute resolution or mediation, and this would provide that. So we would like to see it used in courts nationwide.

I would also, I mean, this is hopeful, but it would be nice if people just thought to “Judicate it”...

Vickie Wood (41:19)
Yeah.

Valerie Clemen (41:19)
when they have disputes, to go there first. You see people arguing on Nextdoor. Sometimes you think we should have guerrilla marketing and just go in there and say, “Just Judicate it.” Like, there’s so many...

Vickie Wood (41:32)
I don’t know if Nextdoor is national, but it’s this neighborhood platform. And yeah, so...

Greg Lambert (41:35)
It is. You know. No, we all...

Valerie Clemen (41:36)
It is.

Greg Lambert (41:38)
know. “Was that fireworks or gunshots?” is the most popular...

Valerie Clemen (41:41)
Yes, exactly. Exactly. “Does anyone...”

Vickie Wood (41:46)
Yeah.

Valerie Clemen (41:44)
“recognize this car?”

Greg Lambert (41:46)
Ha ha.

Valerie Clemen (41:47)
But people have disputes, and it’s really hard to get people talking to each other. Oftentimes, when people start talking to each other, they resolve things. It’s that initial step, because it’s uncomfortable, it’s adversarial. But once people do it, they can often resolve a lot of issues.

So the idea that people could use this as a tool to help get things started, it’s really hard to walk up to someone...

Marlene Gebauer (42:12)
Okay.

Valerie Clemen (42:13)
and confront them and talk about something, even if you have it simmering. I mean, people have a hard time with that with family members. You’ve had this thing simmering forever, but you can’t address it.

This is a tool that helps you be a little protected but still share your thoughts and feelings and resolve disputes. The idea that people could use it for all kinds of disputes, I mean, it is a very civil way to resolve disputes and quash things, and not have things linger for a long time.

But obviously, in the legal sense, it would be nice when people do have disputes to the level where they’re ready to file a lawsuit or do something more, that they have a way to either solve it beforehand or try to solve it once the lawsuit has started. They don’t have attorneys helping them in a lot of these cases. They’re probably using Claude or ChatGPT to help them with their case. So this is another resource to help them resolve it instead of keep fighting.

Vickie Wood (43:09)
Yeah, or have it escalate to something more than it ever needed to be. And just to tack on to that, I think that’s exactly right. A lot of times it’s barriers to communication, and that’s what Judicaid is trying to resolve. Whether it’s lack of skill with communication, a power dynamic, or a language barrier, that’s really where Judicaid provides this platform at a very base level.

And hopefully, this is my tagline that I have in our little video, my hope is someday that instead of saying, “Let’s take it to court,” we’ll say, “Let’s just Judicate it” instead. So that’s our hope.

Marlene Gebauer (43:49)
I like it. I like it.

Well, Judge Wood and Valerie Clemen, thank you for joining us and giving us a closer look at Judicaid, at both the opportunities and difficult questions that come with using AI and expanding access to dispute resolution. And thanks to all of you for listening to The Geek in Review. If you enjoyed the show, please share it with a colleague. We’d also love to hear from you on LinkedIn and Substack.

Greg Lambert (44:17)
And Judge Wood and Valerie, where can listeners go and find out more about Judicaid?

Vickie Wood (44:26)
Judicaid.com. We have a lot of information on there, including an instructional video that’s on YouTube, but you can access it from the homepage, along with how-to instructions. We’re on Instagram, Facebook, X, and LinkedIn, but that’s probably the best place to start.

Greg Lambert (44:47)
Soon to be on Nextdoor.

Marlene Gebauer (44:48)
All the places.

Valerie Clemen (44:50)
Yeah.

Vickie Wood (44:51)
I actually am on Nextdoor. That’s true. I haven’t really spent a lot of time on there yet, but yeah...

Marlene Gebauer (44:55)
Ha.

Vickie Wood (44:56)
I’m going there, Greg.

Greg Lambert (44:59)
Awesome, awesome.

Vickie Wood (45:00)
Thank you so much.

Marlene Gebauer (45:02)
And as always, the music you hear is from Jerry David DeCicca. Thank you, Jerry. Goodbye, everybody.