I knew this was going to happen sooner or later, we all knew it was going to happen, but most of the experiments i ran in 2025, using ChatGPT, Claude for conversation and asset creation, using Make.com and Microsoft Copilot for work flow automation produced less than stellar results. They showed promise but that promise seemed a long way off. Hype cycle versus reality.

This is no longer true, not by a long shot.

On the 25th November 2025 Anthropic released Opus 4.5 a ‘reasoning model’ which is able to use tools and ‘planning’ mode to complete task for you. Developers were the first to notice the productivity boost and they did what developers always do pushed the boundaries and started to use it for more than just dev tasks. They used it organise their file system, connect to browsers to test website they’re building (not creating UI automation script, using your browers on your machine to take actions). Anthropic took note and on the 12th Jan after only 10 days of dev work with Claude code they built and released Claude Cowork. A research project that uses Claude code as a general purpose agent to do non technical work.

I started using it in anger shortly after and what i’ve done with it is nothing short of amazing, exciting and slighlty terrifying.

What have i done with Claude Code/Cowork that i couldn’t do before?

1. I created a skill that will automatically brand any document into the UXC brand
2. I’ve created a command that will read the files in my ‘Meetings to action’ file on my computer to summarise the meetings, put the summary in my project folders and draft a follow up email to the participants
3. I’m able to clients quotes using my estimates sheet, and product service offering data to create brand new presentations, branded and ready to send out.
4. I have created fully working prototype from Figma, and collaborative development to enhance, validate and deploy to a website for my clients to interact with.
5. I created a ux-researcher agent that will try to complete a task on a specified website (using my local browesr) and let our experts user researcher focus on the areas that require most attention rather than doing broad unfocused expert reviews.

My team have been busy too they’ve been using to create passion projects that didn’t have the time for, or capability to delivery without the support of our engineering team.

The blockers to experimentation and discovery are gone, I can spin up projects and assets at the speed of thought. All those ideas i had but couldn’t action are now within my reach.

I l believe the promise that a new wave of entrepreneurship is at our fingertips, bringing with it new opportunities for wealth generation.

Stepping back and assessing this new reality

In May last year I wrote [Do you fit into the delivery team do the future][https://matthewgoddard.net/2025/05/26/do-you-fit-into-the-delivery-team-of-the-future/] at the time I envisaged that would be a few years away. I was wrong. It’s now.

My vision is that domain experts would work with AI Engineers to frame the problem space, define the architecture and get AI Agents to so the heavy lifting.

This vision is still true, AI and Agents are great, but the still need people with domain expertise to give them the right context to ensure what’s being built is desirable, secure and aligned to business and customer needs.

You only have to look at the graveyard that is the great vibe coding experiments of 2025 to see this is true.

In the hand of inexperienced practitioners you risk producing slop, but the hands of experts you create something magical.

We’ve seen this before, we saw it in the dotcom boom, in the professionalisation of the digital industries.

Are you heeding the call?

Now’s the time to act, to learn and to embrace this new way of working.

There are two waves of transformation that are happening right now and only a small part of that is focused on technology.

As BCG says, AI transformation is 10% Tech, 20% Process and 70% People focused.

There four levels of AI adoption:

1. The Bystanders
2. The Conversationalist (Consultant and Collaborator)
3. The Automator
4. The Orchestrator

Transformation 1. Moving from Bystander to a collaborative conversationalist, this is about personal productivity gains and will be mandatory for all future work engagements.

Transformation 2. Moving from Conversationalist to Orchestrator, this is about creating organisational efficiencies and systematically integrating AI into our core business processes.

This transformation requires domain experts and AI engineers to work together to identify, and implement new, ai enabled workflow to unlock capacity and drive new business growth.

This is the scary next step, this is where the trust deficit rears its ugly face, where the spector of AI related layoffs looms large.

There has been too much investment in this new era of AI for this to simply go away, transformation is not an option it’s mandatory, just as the digital transformation of the past 30 were.

Are you ready to take this seriously and start the transformation require to compete, not in five years or ten years, but now?

AI/Agents are now ready to go work, the companies that embrace this transformation will be the companies that thrive in the next era of work.

If AI can research, write, code, analyse, summarise, transcribe and translate… what’s left that makes you valuable?

I think there are two things, neither are new, but both matter more now than they did even a year ago.

• First is how you think
• Second is how you work

Let’s take the first, the thinking bit.
There’s no substitute for clarity of thought. For being able to listen to a messy brief, ask a better question, spot a contradiction, or frame a problem in a way that makes the answer easier to find.

A lot of what we call intelligence at work comes down to this. Not how much you know, or how fast you can write a slide, but how you frame the work in the first place. Most of us have spent years developing this skill without giving it a name. Some of us never do. But it’s what clients and colleagues are really responding to when they say someone is “sharp”.

This kind of thinking is what gives direction to the tools. Without it, you’re just decorating ideas you don’t understand with language that sounds clever.

Then there’s how you work. And this is where the AI part gets practical.

Building your AI toolkit

Right now, the best way to improve your thinking is to enhance your working; specifically, how you work with the tools, how you shape your process to take advantage of what’s now possible. That doesn’t mean waiting for plug-and-play agents to magically change how your team works. It means doing the groundwork yourself.

That means building what I’ve started calling your AI toolkit.

It’s not a platform. It’s not something you buy. It’s something you grow. You make it as you go.
It has three parts, and they’re not all equally important.

Prompt Library

Start here. This is the single most useful thing you can develop today. It’s your personal set of reusable, tested instructions for getting better outcomes from AI models. Good prompts act like tools in a shed. You don’t always use them the same way, but they give you a head start. Especially if you’ve annotated them, tweaked them, tried them in different contexts.

The real kicker here is your context library. This is the stuff only you have, your ways of working, your client materials, your domain knowledge, your frameworks. There is no secret sauce here, yet this is how two people using the same model can get radically different results.

Workflows

Once you’ve got decent prompts, you can start joining them up. That’s where workflows come in. A workflow is just a process with a purpose. Most of us have dozens of these: writing briefs, reviewing work, onboarding clients, chasing updates, summarising notes.

If you can map a workflow and spot the parts where judgment is low but labour is high, you can start to layer in AI to do the heavy lifting. These become repeatable. And suddenly you’ve gone from “experimenting with prompts” to “working differently.”

Agents

This one’s a little further off for most people, but it’s worth understanding early. Agents are the autonomous bits. They’ll be the ones that pick up your prompts and context, run your workflows, and hand you results. Some of this is already here, though it’s still a bit clunky. But the idea is sound. You don’t want to be doing routine tasks if an agent can do it faster, cheaper, and more reliably.

The difference, again, won’t be in the tool itself; it’ll be in the thinking behind it. Your agent will only be as good as the prompt library and context library you’ve given it.

In summary

If you’re reading this and you’re in a role where you need to produce good work, quickly, and in ways that matter to other people, this is probably worth getting your head around.

In a year or two, you’ll be working alongside people who’ve already figured this stuff out.

And when the playing field levels, what’s going to set you apart isn’t your access to the tools. It’s how well you use them and how clearly you think.

I’ve been experimenting with prompt chaining because I needed to get something done that a single prompt just couldn’t handle.

I was trying to pull together a draft proposal from a bunch of messy notes. Usually, that’s fine, but the jump from “here are my notes” to “here’s a coherent, human-sounding doc I can share” is where most of the value (and the pain) lives.

Breaking that work into a few smaller steps, asking for a summary first, then pulling themes, then drafting each section one at a time, helps A LOT.

The final doc still needed a pass, obviously. But I wasn’t starting from a blank page and the thinking behind it felt more deliberate.

What actually is prompt chaining?

Prompt chaining is just passing the baton from one prompt to another. Instead of trying to get the AI to do a big job in one go, you break it into parts. Like:

– First: “Summarise these notes.”
– Then: “Group the points into themes.”
– Then: “Write an intro paragraph based on that.”
– And finally: “Make it less dry.”

You can string those steps together like a little assembly line. You can even run a few steps in parallel, then stitch the results back together at the end.

The great this you can do this in your preferred chat tool by copying and pasting from one prompt to the next, OR you can create a workflow in Copilot Studio, Zapier, Make.com
Where I’ve used it

A few places where this has helped:

* Discovery synthesis: Summarising interview notes, extracting user goals, and then mapping those to opportunity areas.
* Persona building: clustering behaviour patterns, naming them, then generating narratives from the clusters.
* Proposal work: rough first draft from a brief, then tone-of-voice adjustment, then layering in case studies.
* Workshop prep: turning challenges into HMWs, generating ideas, ranking them—all as separate turns.

It’s not magic. But it’s faster than doing it manually, and it still leaves room for judgment at the end.

### What to watch out for

There are some obvious traps.

If the first step is off, the rest usually is too. And if you’re chaining loads of prompts together without stopping to check, you can end up over-engineering something that would’ve been quicker to write yourself.

Sometimes it’s better to just write the damn thing.

Why it’s worth learning anyway

Most people trying to use AI in their work start with the question “How can I get it to do this entire task for me?” which is rarely the right question. A better one might be: “Where in this task do I slow down or get stuck?”

Then you can ask AI to help with that bit. Chain a few of those together, and you’ve got a process that actually works.

Not every job needs a chain. But some definitely do.

This picture my wife made to remember the pets we’ve lost over the year breaks my heart. Not just for the loss it represents, I still miss each of these animals.

It breaks my heart because I was with each one of them when they were put to sleep. I stroked their heads as they fell unconscious and took their last breaths. Each one of them entered the vets on a particular day trusting I would care for them and ultimately bring them home like I did every other time.

I know with each of them I did the right thing, I always say we’re more humain to our pets then we are to ourselves and our loved. With us we let disease and injury ravage us until our bodies can no longer cope and we die.

I know we did the right thing but every time I leave the vets I’m plagued by the thought that in the end I betrayed their trust, I let them down. I should have made them better, I should have brought them home.

Loss and guilt always, and it never seems to go. It taints all my memories of the pets I’ve lost. The last memory with them is the first that comes to my mind – however fleetingly – every time I think of them.

We have Banjo left, after that I think I’m done. I’m not sure how much more loss and guilt I can take.

Today is a sad day!

In December 2024 Polly stated to show symptoms of the Constrictive myelopathy coming back. In January 2025 we took her back to the wonderful team at Lumbry Park. Sadly it was discovered that scar tissue had been forming around the original surgery site was now pressing on her spinal cord, there was nothing that could be done.

Since then Polly has found it increasingly difficult to walk and had lost control of her bowl, now most recently, her bladder.

Every red line we drew for her, we walked over, but last Wednesday, 2025-06-18, I took her to vets for a check up and to discuss her current state and confirmed what we already knew, that the decision we had been putting off needed to be taken.

The decision to have Polly put to sleep has been quite frankly one of the hardest decisions I’ve ever had to make. She’s perfectly fine from the middle of her back forward, and because her nerves are being compressed, she is not in any pain. These facts make it easy to fool yourself. Still, the loss of control over her bladder and the fact that she can’t use her back legs at all mean that she is not able to go for walks, and she is very susceptible to secondary infection and a myriad of other complications.

So yes, today is a very sad day but also i’m thankful for the additional two years we had, and the 7 years Polly spent with us. I love her very much and will miss her deeply.

An article on ‘vibe designing’ triggered a great conversation in our team. It paints a familiar picture: product engineers taking on design work to move faster and iterate in code.

With AI tools, this crossover is becoming easier. A designer can generate code. A developer can generate layouts. But just because it’s easier to emulate another discipline doesn’t mean you’re doing it well.

These tools present ideas with such clarity and confidence that it’s easy to mistake them for complete solutions. That confidence can be misleading. It can make something look like it’s solving a business goal, serving user needs, and sitting on solid technical architecture. But without the foundational knowledge, it’s surface-level.

This feels like another version of a tools-first mindset: “We’ve got something new. How do we keep doing what we’ve always done, just faster?”

A better question might be: How should we rethink our teams and our ways of working to build better products with these tools in the mix?

I wrote about this tension before:

• Designers writing code, developers designing
• Speculative take on the future of delivery teams

Curious how others are handling this. Are your teams redrawing the lines, or working harder to keep them in place?

People expect a lot from AI these days, sometimes too much. Sure, it’s incredible at speeding things up but AI isn’t great at finessing the final details yet.

Its strength is getting you a running start, not a polished finish. Your goal is to get your design, code, presentation, or doc far enough that you’ve saved yourself meaningful time. The value, at the moment, comes from that head start.

But how do you know when it’s time to stop?

There are a couple of signals:
* You’re going in circles: You prompt for a change, it kind of happens, but then you’re back tweaking or narrowing the same thing over and over.
* You get the nagging feeling: “It’d be easier to start from scratch than keep fixing this.”

That’s your subconscious telling you: It’s time to stop. Remember, the goal is to accelerate your work, not to finish it.

Only you can define what “done” looks like. AI can’t read your mind. If you keep pushing it, you’ll fall into the sunk cost fallacy, believing you must finish your task using AI because you’ve already spent so much time with it.

For example, if you’re creating a prototype for a dev and you’re getting stuck, stop and document the changes that are still needed. You’ve given them an accelerator, now help them understand what needs to happen next. You’ve invested hours with AI to save days.

Think about it for a moment, you’ve invested hours with AI to save days. How long would it have taken you to create a code-based prototype? Could you have even done that before? You have accelerated the whole process. Sure, what you did isn’t 100% perfect, but you’re much further down the road than you would have been.

Don’t get trapped. Stop. Use whatever other tool or skill you need to cross the finish line.

I had the pleasure of attending one of Allie K. Miller‘s AI First Mindset masterclasses. If you get the chance to attend one, I highly recommend it. Allie is engaging, clear, and incredibly knowledgeable about her craft.

One thing she said that really resonated with me, I paraphrase

“I can give you examples [of specific workflows], but if you don’t learn the fundamental principles, you won’t be able to scale.”

She shared plenty of great examples, but what really stood out were the four 4 AI interaction modes she presented:

AI as Microtasker: Handling discrete, one-off tasks
• Searching for information
• Creating content, marketing copy, and emails
• Generating simple images and videos
• Converting data formats
• Fixing isolated errors

AI as Copilot: Collaborating in real time
• Writing documents
• Coding pipelines
• Brainstorming ideas
• Planning a trip together

AI as Delegate: Taking on complex goals
• Research projects
• Monitoring systems and summarising
• Purchasing and comparing items
• Generating candidate lists

AI as Teammate: Transforming teams and systems
• Enhancing team workflows
• Participating in meetings, flagging issues
• Cross-functional project management
• Consolidating enterprise-wide KPIs

As someone who sees AI as a tool, I focus on the people using it rather than AI as an entity. That’s why I was pleased to see the overlap between Allie’s AI interaction modes and my own user archetypes (Bystander, Conversationalist, Automator, Orchestrator)

You can read more about these archetypes in my article here: https://lnkd.in/eDizXgDF

Here’s how they map to Allie’s AI contexts:
• The Bystander: AI as Microtasker (basic)
• The Conversationalist: AI as Microtasker (basic) and AI as Copilot (advanced)
• The Automator: AI as Copilot (basic) and AI as Delegate (advanced)
• The Orchestrator: AI as Delegate (basic) and AI as Teammate (advanced)

This fusion of frameworks creates a richer understanding of the spread of AI usage across archetypes.

One key realisation:

It’s easy to gauge how AI-first your business is by looking at the dominant archetypes in your organisation.
• Bystanders and Conversationalists tend to focus inward: How does this affect me? How can I improve myself?
• Automators and Orchestrators tend to focus outward: How can this improve my team and my organisation?

From an organisation’s perspective, we want to nurture more Automators and Orchestrators because they are the ones who drive real value.

Curious about how to transition your team from Bystanders to Orchestrators? Let’s chat. Follow me for more insights, or give me a call, and I’d be happy to discuss how we can support you.

The AI landscape is rapidly changing, and as business leaders, so must we.

Our latest talk breaks down the AI journey into manageable stages:

• Bystander.
• Conversationlist.
• Automator.
• Orchestrator.

Whether you’re just starting or looking to revolutionize how your team works, there’s a path forward. Learn how to transform fear into action and curiosity into expertise with our step-by-step insights. Listen in and find out where you and your team are on the AI-adoption curve.

#AILeadership #FutureOfWork #Technology

AI-shortened transcript

Nick: One of the things that we know people are struggling with when it comes to AI — because they’re telling us — is that it seems so big, so all-encompassing, that it’s really hard to think about from a strategic point of view. What you should be doing, what your team should be doing, what your business should be doing. It’s a bit “jab, jab, right hook,” isn’t it?

Matt: Yeah.

Nick: You published a really interesting post on LinkedIn last week that I think gives us a useful framework for thinking about where we are — each of us individually and from a business point of view — on that AI adoption curve. And I thought it would be useful just to talk through those levels and see if we can help some people out.

Matt: Absolutely. I think it’s worth saying, just for context — Nick, you and I have spoken about this a lot — and I want to reiterate a little bit, because I think this is really important. I’m years old now. I’m living through a time where I’m seeing, genuinely, for the first time, a real worry and threat to the way I’ve been able to do my work for such a long time.

You know, I was a developer, I’m a user experience person, I’ve done a lot of product strategy — I’m not a junior person in the digital space at all. But every time I look at LinkedIn, Instagram, TikTok, or Twitter, all I’m seeing again and again are these messages saying AI’s coming. It can take your job, it can do this, that, and the other. And I genuinely think to myself, “Oh my — crikey — how am I going to adapt to this?”

It’s scary, right? I feel fear.

Nick: Yeah.

Leaning Into AI

Matt: And as part of that fear, what I’ve been doing — as you know, what we’ve been doing together actually — is trying to lean into AI. Practical lessons that we, as older statesmen of the digital…

Nick: Some of us older than others, Matt.

Matt: …so that we can start to understand the wheat from the chaff, the signal from the noise. Because it’s not obvious. If you go onto Google or YouTube and ask, “What are the five most important impacts of AI right now?”, you’re going to get videos about “vibe,” or how to use this niche tool some dude has vibe-coded — an AI tool that will, say, translate your company accounts. But you know nothing about the provenance regarding security and all that.

There is so much noise about what’s good and bad. It’s really hard to tell the difference. So we’ve been leaning into that from experience. We know how to assess good from bad in terms of…

Nick: Yeah, practical applications.

Matt: …practical application of these things. And what occurred to me was that I haven’t been able to find a very good framework for understanding where I am on my AI adoption journey. For example, if I’m just using ChatGPT to ask it questions — basically like a Google search — what does that mean?

There’s a woman called Allie Miller who’s one of the top voices on AI on LinkedIn. Really great — her webinars are amazing. But her work seems targeted very much to younger enthusiasts — millennials and Gen Zs. I’m a Gen Xer.

Her posts are great, but very much geared toward that audience. So I kept thinking: how do I understand my journey on this? She talks about an “AI-first mentality” — about using AI to change the way you work, not just to use AI. What does that mean?

If I’m using ChatGPT to analyze a spreadsheet and understand sales trends in the US, and ask it how tariffs will affect those sales — that’s still me using the tool to do what I’d normally do manually. I’m saving time, sure. But that’s not AI-first.

AI-first would be: a workflow that sees Donald Trump announce 50% tariffs on steel, then asks: “Do we work in the steel industry?” Yes. “OK, I’ll look at our sales forecast and tell the heads of department what the impact is.” That’s AI-first — it’s proactive, not reactive.

Nick: And it would be running in the background all the time. Not triggered — just alerting you.

Matt: Exactly. So what I’ve been trying to get my head around is: what are the gates or behavioral markers at each stage, so I know I’ve moved from being a basic user to being an AI-first individual?

Nick: Yeah. And it’s not just you, is it? There are lots of people working in companies — doing big work for big clients — who aren’t stringing together no-code tools. It’s the complex work of business.

Matt: Right. And that’s what really got me thinking. There are lots of videos about how to use AI as a content accelerator. But there’s not much out there about how to use AI to be a better team leader or line manager. Those are harder use cases. They’re specific to most organizations.

The Four AI Archetypes

Matt: So we’ve got these four archetypes I call AI User Archetypes: Bystanders, Conversationalists, Automators, and Orchestrators.

Nick: OK.

Matt: Originally there were only three — Conversationalist, Automator, and Orchestrator. But I had a conversation with someone on LinkedIn who said, “I’m not interested in AI; it’s not solving any problem for me.” And I realized I’d missed out the archetype I myself had once been — what I call the Bystander.

Nick: Yeah.

1. Bystanders

Matt: At a high level, a Bystander is standing at the edge of the AI revolution, watching from a distance. They’ve heard of AI, maybe tried a chatbot, but it hasn’t stuck. They don’t think it’s relevant to their work. They might think it’s overhyped or unreliable. They feel a mix of curiosity, skepticism, and fear. What they need is clearer use cases — right now, AI feels like it’s for someone else.

2. Conversationalists

Matt: A Conversationalist uses tools like ChatGPT, Claude, Gemini, or Copilot to ask questions, draft emails, brainstorm ideas. It’s one-on-one interactions in a chat window. The AI is a sparring partner or research assistant — it’s not embedded in any workflow.

Nick: Yeah, it’s separate.

Matt: Exactly. And the landscape is muddied — there are ads claiming AI can automate your life, but they’re promising things that ChatGPT alone can’t actually do. So, Conversationalists need to understand that these tools are like a keen but inexperienced junior employee — they might confabulate because they’re trying to give you an answer.

Nick: Yeah. Confidence is misleading — AI seems confident, and we fall for that.

3. Automators

Matt: The next level is Automators — people who design workflows where AI is a step in the process. We recently built one for a hotel chain to generate SOPs from vendor training videos. You use AI to summarize, translate, clean data, or serve as an FAQ bot. It’s not just chat — it’s embedded in tools and flows, often using Copilot Studio, Make.com, or Zapier.

For example: add a line in a Google Sheet to trigger ChatGPT to create a draft invoice in Xero, then alert you via Slack.

Nick: Yeah, that makes sense.

4. Orchestrators

Matt: At the Orchestrator level, you’ve got AI agents working proactively. I have one in my inbox — it watches for action-required emails and adds tasks to my to-do list automatically. It has a prompt like “You are a first-class personal assistant.” It checks email and uses tools to capture the task.

Over time, you build many agents — maybe a coding agent to create tools based on your brief. You’re not doing the work; you’re ensuring it gets done right.

Senior leaders need to move their teams from Bystanders and Conversationalists to Automators and Orchestrators. Because what AI can’t do well — and may never do — is provide the creative spark to ask: why are we doing this? That’s the domain of humans.

Nick: Yes — experience, creativity, empathy. That’s the rarefied space where we create value.

The Future of Work

Matt: There’s worry about what happens to junior roles. Anthropic recently said 50% of junior roles could disappear in the next three years. That’s scary.

Nick: Yeah.

Matt: But I think managing agents, workflows, and automations could create a new area of work. It might not be the same tasks, but new kinds of roles will emerge.

Nick: Agility and hard work will be the differentiators. It’s like writing — yes, AI can write, but thinking is what really matters. If you just use AI to write, you’re the same as everyone else doing that. But if you push harder, you can do more — coding, designing, prototyping. You can go farther, faster.

Matt: Totally agree. There’s this negative vibe-coding label out there. But the people using these tools are learning — just in a different way. They’ll be ahead of us if we don’t adapt. Leaders must remind their teams: if you’re taking 10 days to do something that AI can help someone else do in two hours, you’re falling behind.

Nick: Exactly. If you’re scared, you can run, freeze, or dive in. Freezing is the problem. Knowing where you are in this framework tells you the next step. And moving forward is the only option.

Matt: Couldn’t agree more.

Nick: Cool. Thanks mate. Speak to you again soon.

Matt: Will do. See you later.

I tried going AI-first this week. Spoiler: it’s brilliant… until it isn’t.

I sat down with Nick Warren to unpack what happens when you actually try to run your daily workflow through AI.

No filters. No hype. Just two humans navigating hallucinations, rabbit holes, and the occasional win that saves a week’s work.In this chat:

• Why I treat ChatGPT like an overly confident intern (mine’s called Wilson).
• The moment it lied to me about Flask.
• When it builds you a webpage… that looks nothing like your website.
• Why voice input is finally working — and what it means for how we work next.
• And yes, why Git still saves lives.

If you’re serious about working with AI (not just tweeting about it), give this a listen. (We’ll try to improve the audio next time!)#AI #ChatGPT #TechWorkflow #ProductivityTools #DevLife #PromptEngineering

Transcript

Nick: [00:00:00] So Matt, I’ve got a confession to make. You know, last time we recorded, we were talking about AI. But when we finished, we weren’t sure whether or not to post it, because AI is such a big thing and we didn’t reach a ton of conclusions.

It’s not like the usual time we talk where —

Matt: Yeah, that’s right.

Nick: — you’re unloading your wisdom on us. But when we decided we were going to put it out, we decided to just publish the video with the transcript. And I said to you, yeah, that’s great. But I’m going to have to do quite a lot of work on the transcript because — if you’ve ever seen a transcript — from about shifting the default behaviours… And I think, you know —

Matt: Yeah.

Nick: Last time we were talking about how, you know, if you are just starting out, or maybe now you are 10 years old and obviously you’re going to grow up an AI native in the way that my son grew up a digital native. But for the rest of us who are older, or for businesses, one of the key questions we need to ask [00:01:00] ourselves when we’re doing any job — particularly if it’s essentially busy work like that — is: how should I use AI? How can I use AI to get this done quicker and move on with the things I can’t do. Does that ring any bells for you?

Matt: Yeah. Oh, absolutely. Absolutely. Do you know, it’s such an interesting point. Funny enough, yesterday or the day before, I was pulling together a list of use cases — like, how might you use AI to accelerate your daily work life?

I was Googling and going on YouTube saying, “Give me some use cases.” And you know what? There is just so much noise. On YouTube, there is so much. I’m trying to find the most influential people and what they’re saying about how to use AI.

Then there’s [00:02:00] this whole mix between how to use tools like Claude and ChatGPT versus workflow automation and stuff like that. By the end of it, my brain was just like: what the hell?

Where I’ve landed on this is taking an “AI first” approach. Everything I want to do, I’ll try with AI first. If it doesn’t improve things, I stop. If it does, then great.

Nick: Yeah.

Matt: Yesterday I had a list of tasks. One of them was to produce some copy for a welcome screen for a questionnaire survey, and then a thank-you screen. I used ChatGPT — who I’ve named Wilson. I gave him the context and he produced the copy. I added it into the system, reviewed it, moved it forward.

Then I needed to create some content for a web page about AI skills — how we’re using AI, what the skills and capabilities are. I did that too. Then I thought, okay, I’ll get it to look on the web to find out what other pages are like, and not just create the copy but the whole site.

Nick: Ah.

Matt: That wasn’t nearly as successful. [00:04:00] Clearly there’s a challenge in finding the right tool for that. Previously I’d used Vercel’s V0 tool to build web pages. This time I was just using ChatGPT.

Nick: You were generating the HTML, were you?

Matt: Yeah. I said, “This is our current website. Look at the landing pages and create a new one using the AI content.”

It wasn’t even close. It didn’t resemble our existing pages.

Nick: So —

Matt: But that’s good, because I’m understanding the limits.

Nick: Right. Because what we’re really talking about is how we’re pushing AI into the workflow. And that does mean you’re right at the edge of what it can do.

Here’s my prediction: even though AI failed yesterday, right up to the point it failed, it was very confident it could do the job.

Matt: Yeah.

Nick: That’s my experience. When you push it to the limit, it says, “No problem!” And then… you sit there for half an hour.

Matt: I —

Nick: Because it can’t actually cope yet. And sometimes it’s about context length. [00:05:00] That’s what I hit most often.

Matt: Yeah.

Nick: If you’re listening and you’re not sure what context length is: it’s how much data the LLM can process before responding. I might feed it a ton of links and say, “Let’s analyse these web pages.” But sometimes that’s just too much.

Still, I kind of like its unbridled enthusiasm up to a point.

Matt: Yeah.

Nick: And then I tell it to get its act together. Have you seen the thing OpenAI has done — where ChatGPT offers to describe you based on your conversations?

Matt: Oh yeah, I saw that on LinkedIn. Then I tried the prompt myself, and was shocked. The thing is, it feels a lot like reading a horoscope.

Nick: Right. It could apply to anyone.

Matt: Yeah. But then again, people post them on social media like it’s gospel. Did you try it?

Nick: I didn’t post it, but yeah. The prompt just popped up in the app. It basically said, “Want to know what I think of you?”

Matt: Really?

Nick: It was flattering, as you’d expect. But then I asked something more useful: “What should I work on, based on your experience of me over the past year?”

Matt: Right. Okay.

Nick: Because sure, it’s nice being told what you’re good at — but at this stage, I know that. What I want to know is where I can improve. That’s how I framed it.

Matt: That’s interesting, especially when we’re talking about confidence. AI is so confident it can do something — until it can’t. You really have to be on your toes.

For example, this morning I was using Copilot in Visual Studio Code to create a web interface. I used Python, so it picked Flask — a web framework.

It gave me instructions, but told me to run the app using python filename, which is wrong. For Flask you run it using flask run. It told me the wrong thing multiple times.

In the end I corrected it, and it updated the README file accordingly. But the point is: it was so confident.

Unless you know better — unless you know the limitations and hallucinations — you might never question it. [00:09:00] Your ability to use AI effectively is limited by your own knowledge.

Nick: Yeah.

Matt: I’ve had that loads of times, especially with code. It’s only through testing and validating that I’ve caught its mistakes. Maybe that’s less of a thing with content?

Nick: Creating text is what it’s best at — for now. But I’ve seen the same thing happen in code, spinning up a Hugo site, PHP, terminal work — stuff I don’t normally do.

Same thing with Python. There have been times I’ve asked it for something, and because I didn’t know better, I ended up wasting time going down a rabbit hole.

Matt: But that’s the thing. You might waste half an hour, but save yourself a week’s worth of effort overall. So that’s not a bad trade-off.

Nick: I think that’s true — as long as you get somewhere by the end of that half hour. There were times where I didn’t.

One example: we were messing with a Hugo template. Of course, being me, I wanted it to be slightly different from the designer’s version. I kept asking ChatGPT to make changes — “Now do this, now do that” — and eventually it got so mangled that it was easier to start again.

Matt: Yeah. That’s where best practices matter. From a development standpoint, AI is only an accelerator within good process. If it’s changing your code a lot, you need Git.

Nick: Yeah.

Matt: Because asking ChatGPT to undo changes is hard. It doesn’t track what it did well enough to roll things back. But if you’re using Git, you can revert the changes.

A lot of people using AI don’t have that developer background. They don’t know how to make their lives easier in that way.

There’s a bigger conversation here about experienced professionals versus new users, and job loss versus productivity.

There are so many peripheral tasks in every job — software development, writing, UX — that AI doesn’t cover yet. That makes it hard to replace people wholesale.

Nick: Yeah. And on that Hugo project, we did use Git the second time around — because I’d asked those questions. “How could I avoid this problem next time?”

Matt: Right.

Nick: It’s all about building a better process. Having a checklist. Writing it down. Improving it each time you hit a snag.

Eventually it becomes second nature.

One other thing I wanted to tell you — this week I’ve started talking to ChatGPT.

Matt: Really?

Nick: Yeah. You mentioned naming it Wilson. I don’t know if we talked about this before, but I think the big shift will be when we get personal AI — something that really knows us.

But for now, what I’ve been doing is hitting the dictation button on my Mac and speaking my prompts.

Matt: Oh, so you mean speech to text?

Nick: Yeah. I just tap the button, speak what I want, hit enter.

It’s part of smoothing the process. A lot of prompts aren’t programmatic. They’re casual — “I want this, not that.” And I was spending too much time typing. It’s faster to speak.

If you work somewhere you can talk out loud, try it. We’ll all be talking to our computers soon. Better to get used to it.

Matt: You know, that’s so interesting. Years ago, Microsoft did some 3D modelling thing for Xbox. The idea was gestural and voice interfaces would change UX.

In UX theory, major changes come from input shifts — like the mouse and GUI revolutionised computing.

People have talked about voice for years, but it didn’t work well — speech-to-text was rubbish. Not anymore.

Nick: Yeah.

Matt: But the challenge now is the environment. Imagine an office where everyone’s talking to their computer.

Nick: True.

Matt: You’ve got people in meetings, noise bleeding through, noise suppression not working — it’s tricky in enterprise environments.

But like you said, for those who can talk out loud, now’s the time. I hadn’t thought about this in ages, but the dream of natural language input is actually here.

Nick: I got excited about dictation 15 years ago. And again when I was writing novels. But it never worked — detection wasn’t good enough, and I’m a fast typist, so it was easier to type.

Matt: Yeah.

Nick: Now, the detection is much better. And crucially, we’re no longer asking it to create the final output. We’re asking the LLM to interpret what we meant.

Matt: And then generate the final result.

Nick: Exactly. While noise might be an issue for some, I’ve worked in offices where everyone was on the phone. So I think we’ll adapt.

You’ll be shouting to your computer from across the room while making tea: “Hey, do this.”

Matt: And it’s believable. At home, we’ve got an Apple TV and some HomePods. It recognises who’s speaking — my wife, my daughter, me. Shows an icon to indicate who gave the command.

Nick: HomePods.

Matt: That kind of detection is solid now. That’s why AI feels so hard — it’s hard to know how to use it.

Just like with remote work or early websites — people know the potential, but don’t know where to start.

Concrete examples help. Like using voice to get outcomes. Or today — I needed to write a case study for a client. We have a consistent format. I asked ChatGPT to interview me and use my answers to write it.

I didn’t need to write it at all. It just needed to extract info from me and deliver.

Nick: I agree. We ended the last episode saying people should just use it. Hopefully this one gives you another nudge in that direction. Make AI your default.

Matt: Yeah. Ask: how can I use AI to do this job? And if you’re working alone — try talking to your computer.

Nick: One trick is to imagine an eager assistant. How would you explain what you want done? That’s how to think about prompts.

For example, I needed that case study written. An assistant wouldn’t know everything — so I’d ask them to interview me. That’s what you tell AI to do.

Prompt engineering matters too. But if you can get it to ask clarifying questions, that helps.

Two tips: 1) Imagine you’re briefing an assistant. 2) Encourage clarifying questions.

Matt: Great tips.

Nick: One last thing: if you’re just using ChatGPT in a browser, try the app. On my Mac it’s right there at my fingertips.

Matt: Maybe next time we talk prompt engineering.

Nick: Yeah, I agree. It’s the first step toward levelling up — from typing random sentences to getting great results.

Matt: Cool.

Nick: We’ll speak to you next time.

Matt: Yeah, speak to you.

Nick: Thanks mate. Bye now.