Before I became a teacher, I made my living as a musician. One of the unexpected lessons I learned during those years had nothing to do with playing my instrument.

A drummer I knew, Gene, told a story about a friend who auditioned to play on tour with The Eagles. The drummer arrived for his audition, only to be told the band was running behind. While they worked through vocal harmonies, he was asked to pick up an egg shaker and quietly played along. Eventually they apologized, thanked him for his time, and asked him to come back the next day.

The next day went much the same way. More waiting. More egg shaker. Still no opportunity to sit behind the drum kit.

When they asked him to come back a third time, he had to be wondering what was going on. Instead of another delayed audition, they simply offered him the gig. Confused, he asked why they had never actually asked him to play the drums.

The answer has stayed with me for years.

“We already knew you could play, or you wouldn’t have been here. We just wanted to know what it was like to hang with you, because we’re going to spend a lot of time together.”

Musicians sometimes call this the hang. It’s the idea that talent gets you in the room, but your personality, your humility, and your ability to make other people better determine whether you get invited back.

Quick aside: I’m using a photo of Gene as the featured image for this post. He passed away a few years ago, but I still think about this story often. It’s one of those lessons that extends far beyond music, and I’m grateful he shared it with me.

I’ve found Gene’s lesson is also true in education… schools are incredibly collaborative places. We spend hours planning together, troubleshooting problems, organizing events, and simply navigating the unpredictable rhythm of a school day. There are certainly moments where expertise matters. I want my colleagues to be thoughtful educators and skilled professionals.

But after a certain point, competence becomes the baseline. What separates people is whether they’re enjoyable to work alongside. Are they curious? Generous with their time? Calm when things don’t go according to plan? Do they leave the room better than they found it?

I still share this story with students who are thinking about pursuing music professionally. They expect the lesson to be about practicing more or becoming a better performer. Instead, it’s about becoming someone people genuinely want to spend time with.

As it turns out, that may be one of the most valuable professional skills any of us can develop.

While cleaning up old WordPress sites during my recent website migration, I stumbled across a page of notes from an NAIS roundtable I attended in early 2021. At the time, every independent school was trying to answer the same question: How do we preserve our traditions when we can’t gather together?

The notes were meant only for me, but reading them five years later, I was struck by how relevant one idea still feels: Constraint breeds creativity.

The facilitator challenged us to stop asking, “How do we recreate this event online?” Instead, we were asked to begin with a different question: What is the core purpose of this tradition? Design from there.

That shift in thinking unlocked some remarkable ideas.

Our eighth grade musical became a green-screen puppet production. The art show transformed into a virtual gallery featuring every student. Weekly assemblies evolved into edited video broadcasts with guest singers, jokes, and announcements. Student lunch clubs moved online and remained student-led. Outdoor education became faculty-led virtual field trips. Buddies became pen pals, read-alouds, and student-created activities.

None of these were perfect replacements. They weren’t trying to be. They were designed around the purpose of each tradition rather than its familiar format.

One of my favorite examples was our eighth grade recognition ceremony. Instead of one large event with hundreds of attendees, faculty members met individually with each graduate and their family on campus to read their speech in person. Those speeches were also recorded and edited into a premiere video that every family could watch together. Looking back, it wasn’t simply a compromise—it created a different kind of intimacy that the traditional ceremony never had.

Of course, many of these traditions eventually returned to their original form. Most should have. There’s something irreplaceable about gathering together in person.

But I hope we don’t lose the broader lesson.

When we’re willing to ask why a tradition exists before deciding how it should happen, we often discover better ideas than we’d imagined. Sometimes the constraint isn’t the obstacle—it’s the catalyst.

One other lesson is worth holding onto. During those years, we became much better at creating ways for people to participate remotely. Today, most schools have understandably returned to in-person events, but there are still students, family members, and colleagues for whom attending in person isn’t always possible. We now know how to make many of our traditions more accessible. I hope we continue to use those skills whenever we can.

Funny how cleaning out an old blog can uncover ideas that still feel surprisingly current.

For years, whenever students started making videos, the same question inevitably came up: Where do I find music that I’m actually allowed to use?

The good news is that there are more great options than ever. Whether you’re creating a classroom project, a school promo, or just experimenting with video, here are a few of my favorite resources.

If you’re looking for a huge library of royalty-free music, Incompetech is still one of the best places to start. Kevin MacLeod’s collection includes more than 1,300 tracks, and you can filter by genre, mood, tempo, or length to quickly find something that fits your project.

For something a little different, check out Mobygratis. Yes, that Moby. He created the site to make his music available for independent filmmakers, nonprofits, and students. It’s a fantastic collection if you’re looking for something with a more cinematic feel.

If you’re creating videos for YouTube, don’t overlook the YouTube Audio Library. Every track clearly lists its licensing requirements, making it easy for students (and teachers) to stay on the right side of copyright while still finding quality music.

Another newer favorite is Pixabay Music. While many people know Pixabay for free images, its growing music library is excellent for classroom projects and includes a wide variety of modern, high-quality tracks.

One of my favorite hidden gems is a little more personal. My former neighbor and middle school teacher, Steven Wesley Guiles, has hundreds of original songs on SoundCloud. His music wasn’t uploaded as a stock music library, which means you can often find tracks that sound refreshingly different from what everyone else is using. If one of his songs fits your project, reach out—he’s been incredibly supportive of student work.

Finally, if your students have iPads or iPhones, don’t overlook Apple’s Clips app. It’s one of the most underrated video tools for schools. The built-in soundtracks automatically adjust to match the length of your video. Record a 20-second clip? The music fits perfectly. Add another few seconds? The soundtrack seamlessly stretches to match.

I’ve even borrowed that Clips feature for more advanced projects. If I’m editing in Final Cut Pro, I’ll create a temporary Clips video that’s exactly the same length as my finished project, export the soundtrack, and drop it into Final Cut. Instant, perfectly timed background music.

Sometimes the best tools aren’t the newest ones—they’re the ones hiding in plain sight.

When students complete a Google Form to choose a group, how do you put students into groups? And how do you communicate that to students?

Every December, I create 17 (or so) groups for 190 middle school students and 13 more groups for 140 lower school students. Here are my workflows using Google Sheets. I’ll walk you through how I go from getting every students’ top three choices to handing out tickets for the event.

Why should students code?

First things first, be sure you’ve read my post on why every student at Hillbrook spends time coding in a schoolwide event. That’s the why with a few details; this post focuses on the how, particularly in Google Sheets as I organize all the things.

Step 1: Signups

Since middle school and lower school students participate in Everyone Can Code on different days, there are different forms. Check out the MS form and LS form to see how they picked their top 3 choices.

This would be a good time to mention that I make a five minute video for LS students and embed it in the signup form. Keep in mind that younger students need extra support in reading, and extra info since some of them have NO IDEA what these robots and websites are.

Step 2: Organize

This makes way more sense if you’re following along with the workflow. Here’s a copy of the Google Sheet with fake student names. Open it in another tab and keep clicking back and forth as you read.

Google Forms stores entries on a spreadsheet. That Google Sheet has one sheet titled Form Responses. I never edit this data. My firs step is to create a new sheet (plus button at bottom) and customize the headings. The header is the question you ask and “Which session do you want to go to?” isn’t super helpful. So on this new sheet, creatively titled Query, I type new headers in row 1. The cells are simple so far: First Name, Class, Query 1st, Query 2nd, and Query 3rd.

Then, in cell A2, I type =QUERY(‘Form Responses 1’!B2:F160). That pulls the info from that original sheet; every time a student completes the form, it will appear here. We’re about to add some columns, and things get goofy quickly when a form is being completed and you add columns. So, =QUERY and then…

I added a few more columns. Columns like Query 1 are super useful, but they’re filled with way too many words like “Build a Computer with Piper: physical computing on a Raspberry Pi.” I just want it to say Piper. To do that, I created a few more columns: 1st choice, 2nd choice, and 3rd choice. Buckle up, cuz here’s where the formula gets insane. The goal of this is to basically do an automatic Find & Replace for each choice. Note: if you copy and paste this, smart quotes might break the formula. If it doesn’t work for you, either delete the quotes and manually type them or just manually type the whole darn thing. Here’s the code from cell F2:

=IFS( (arrayformula(iferror(search("Jawas",C2)))),"Star Wars",  (arrayformula(iferror(search("aRT",C2)))),"Sphero Art",  (arrayformula(iferror(search("Hackathon",C2)))),"Hackathon",  (arrayformula(iferror(search("code a dance party",C2)))),"Dance Party", (arrayformula(iferror(search("LittleBits",C2)))),"LittleBits", (arrayformula(iferror(search("An Unusual Discovery: Make a Cartoon w/ Code",C2)))),"Unusual Discovery", (arrayformula(iferror(search("Build a Computer with Piper: physical computing on a Raspberry Pi",C2)))),"Piper", (arrayformula(iferror(search("Light-up Holiday Card with Paper Circuitry",C2)))),"Paper Circuitry", (arrayformula(iferror(search("Osmo Coding Kit",C2)))),"Osmo", (arrayformula(iferror(search("Bloxels",C2)))),"Bloxels", (arrayformula(iferror(search("Dash-ketball",C2)))),"Dash-ketball", (arrayformula(iferror(search("Superhero",C2)))),"Superhero", (arrayformula(iferror(search("SonicPi",C2)))),"SonicPi", (arrayformula(iferror(search("Javascript 101",C2)))),"Javascript 101", (arrayformula(iferror(search("Sphero Maze",C2)))),"Sphero Maze", (arrayformula(iferror(search("Sphero Goes Green",C2)))),"Sphero Goes Green", (arrayformula(iferror(search("AI for Oceans",C2)))),"AI for Oceans", (arrayformula(iferror(search("Rock, Paper",C2)))),"Swift Playgrounds", (arrayformula(iferror(search("Chrome Music Labs",C2)))),"Chrome Music Labs", (arrayformula(iferror(search("Makey",C2)))),"MakeyMakey Poetry" )

From there, drag the bottom right corner to copy down the formula. In my case, I dragged it from F2 to F137, and also from F2 to H2, and then down to H137. After this super geeky formula, I finally have easy-to-read titles, automatically generated.

I wanted an even easier way to quickly identify what session a student requested. Enter… Conditional Formatting. I created 13 formats for cell background and text color so that I could quickly see who signed up for what.

At this point, I created another column–Grade level–and did the same IFS formula magic and Conditional Formatting craziness. Here’s what that looked like, starting in cell K2:

=IFS(
(arrayformula(iferror(search(“4″,B2)))),”4th grade”,
(arrayformula(iferror(search(“3″,B2)))),”3rd grade”,
(arrayformula(iferror(search(“2″,B2)))),”2nd grade”,
(arrayformula(iferror(search(“1″,B2)))),”1st grade”)

Easier to understand, right? Based on their answer in column B, I could automatically have an easy-to-read grade level column. Again, because this is on a new sheet I named Query, every time a student fills out the form, this info gets added to it. That’s rad.

I also added Conditional Formatting to column K, my grade level column. This time, it’s much simpler, with 4 shades of grey. The reasoning behind this is that I want to instantly see the age of students when I start creating mixed-age groupings. All of this so far is students sharing their top three picks (out of 13). The next step is for me to create groups and put students from grades 1-4 (or grades 5-8 for MS) into groups for their first session (1-2pm in our case) and their second session (2-3pm).

This new sheet, Query, is awesome, but there’s on big drawback–you can’t sort it. So…

Create a new sheet. I named mine “Work here (won’t pull query)“. Copy the entire sheet named Query and head over to the new sheet and click Edit > Paste Special > Paste Values Only. This no longer updates automatically, but it lets you sort, and that’s a super big deal for the next big step.

Step 3: Create Student Groups

On this new sheet named Work here (won’t pull query), pull down row 1 so that it stays frozen when you sort. Next, go to column F, 1st Choice, grab the drop down triangle, and Sort Sheet A->Z. Your color coding should make it super obvious which sessions are popular for their first choice. You should have a max occupancy in mind for student groups; any time a group is less than max in the 1st choice column, copy and paste those students into the official group… the new fancy column… the one I called Session 1. These students now have a session to attend!

Since the sheet is now sorted by first preference, take a moment and look at grade level. Is the group heterogeneous, or did all the fourth graders sign up for it? Try to get as many students as possible into their first choice without overthinking things or worrying about overbooking a session. Because it’s time for…

Pivot tables! Select the entire page, go to Data > Pivot Table, and select Create New Sheet. Under Row, Click Add, and select Session 1 (check the box Show Totals). Then under Values, Click Add, and select Session 1 (ensure Summarize by CountA is selected). Be nice to yourself and give this sheet a helpful name like Count session 1. This pivot table gives you a real-time count of how many students are in each group. Rad.

Let’s do the same for Session 2… go back to your sheet named Work here (won’t pull query). Select the entire thing and create a Pivot Table. Same as before… Create New Sheet, Row > Add > Session 2, and Values > Add > Session 2. Name this Count Session 2. You should now have two more sheets. This is nice, but can be nicer. QUERY will be our friend again.

Create a new sheet. Name it something like Counts. Manually type the headings: Title, Session 1, and Session 2. Then in cell A2, type =QUERY(‘Count session 1’!A3:B14). If your sheet has a different name, don’t copy and paste this. That query grabs the names of each session and the count of session 1. You’ll need the count of session 2. For that, type =QUERY(‘Count session 2’!B3:B14). For good measure, add some Conditional Formatting to it–when a count goes above your max, make it red. If it’s in the sweet spot (max or a couple below), make it green. If it’s under, make it yellow. Now it’s just a matter of filling all groups to capacity and attempting to give students their top choices. It’s tricky, but between the color coding, easy-to-read session names, and real-time counts on your two sessions, it’s certainly doable.

Step 4: Communicate Sessions to Students

You know where every student should go, but nobody else knows. It’s time to let teachers and students know where to go, and what sessions they’ll be attending.

Middle school student workflow: FormMule

if students have email, let them know their two sessions by email. FormMule

Lower school student workflow: if students don’t have email, print tickets for them. Autocrat and Google Slides

In this episode, I share how I finally moved my websites off Bluehost after more than 20 years. What started as a simple side project with Cloudflare turned into a complete migration of my domains, DNS, and WordPress hosting—and it ended up being much less intimidating than I expected.

Along the way, I talk about why I chose Cloudflare and KnownHost, how ChatGPT helped me work through each step with confidence, and why having an ongoing AI conversation made tackling a project like this feel manageable instead of overwhelming.

If you’ve been putting off a website migration because it feels too risky or too technical, this episode is for you. I’ll walk through what I learned, the mistakes I avoided, and why now might be the perfect time to make the move yourself.

more-text-here-if-needed

I shipped an app! It started with a familiar kind of friction: shared listening that didn’t feel shared. My kids each wanted control of the music. I wanted some version of fairness. And somewhere along the way, my Apple Music Replay stopped reflecting my own taste.

That last part stuck with me more than I expected. Replay is supposed to be a kind of memory—a record of what you actually listen to. But in a shared space, it becomes something else entirely. A blended artifact. A compromise.

So I started asking a simple question: What would it look like if shared listening respected individuals?

That question became NextUp Music.

👉 App Store: https://apps.apple.com/us/app/nextup-music-player/id6757323472

👉 Stoked Logic: https://stokedlogic.com/#apps

The Idea (Before the Code)

The core idea came quickly:

• One song at a time
• One person at a time
• A clear, fair rotation

No queue battles. No invisible algorithms. No background automation deciding what comes next. Just turns.

But ideas like this are easy to state and surprisingly difficult to hold onto. Every step of building an app introduces pressure to add convenience, automation, or “just one more feature.” So I approached this as a series of small sprints—31 in total—not to move fast, but to stay honest.

The First Few Sprints (Setting the Rules)

The early work wasn’t about polish. It was about constraints. (Quick aside for some subtext: I’m hearing a lot of “just tell ChatGPT to build you an app”. These steps below are to help folks understand that there is still A LOT of work that goes into app development. The big shift is simply that you can use agents to write now.)

Sprint 1–3:

Get MusicKit working. Prove that turn-based playback is even possible. Frame listening as a “session” (or a ride), not an endless stream. That shift—from infinite feed to bounded experience—ended up shaping everything.

Sprint 4–6:

Profiles without accounts. Identity without friction. Manual control only. No autoplay. No silent handoffs.

This was an intentional rejection of how most media apps behave. Music doesn’t advance unless someone chooses it. That sounds obvious. It’s not.

Where It Got Interesting (And Harder)

Around the middle sprints, the app started to feel real—and that’s where the tension showed up.

Recent Picks (Sprint 7)

This solved a very practical problem: choosing a song shouldn’t require starting from scratch every time. But it had to stay lightweight. No permanent history. No heavy-handed recommendations.

Search + Context (Sprints 8–9)

Search needed to feel like part of the session, not a separate universe. You’re not browsing all of Apple Music—you’re choosing your next song, right now.

Visual Calm (Sprint 10)

Most media apps are dense. I wanted the opposite. Big, glanceable elements. Especially in a car.

Error Tolerance (Sprint 11)

Real life is messy. Songs get skipped. Connections drop. The app needed to absorb that without breaking the experience or blaming the user. By this point, I realized something: I wasn’t just building features. I was defending a philosophy.

Replay Shield (The Feature That Made It Click)

The moment the app fully made sense was Replay Shield.

Songs played in NextUp don’t affect your Apple Music Replay.

It’s a small sentence. But it solves something surprisingly emotional. If you’re a musician practicing the same track over and over…

If your kids are deep into a very specific phase…

If shared listening is part of your daily life…

Your Replay stops feeling like you. Replay Shield gives that back. And once that was in place, the rest of the app snapped into focus. This wasn’t just about fairness. It was about ownership of listening history.

The Later Sprints (Power Without Breaking Trust)

The final stretch of development was about adding capability without losing the original intent.

Session End + Recap (Sprint 14)

Give the listening experience a sense of closure. A beginning, middle, and end.

Cruise Mode (Sprint 25)

Auto-advance turns—but carefully. Not autoplay. Still bounded. Still human.

Onboarding (Sprint 28)

Shift from explaining buttons to teaching philosophy. What is this app trying to be? Remind (or teach) folks that this is not a typical music player.

Monetization (Sprints 22, 27)

This part mattered more than I expected. I had to decide what should never be behind a paywall. Turn-taking. Core functionality. Those had to remain intact.

Only after that did I layer in optional enhancements.

App Store Readiness (Sprint 31)

Clean up everything. Remove the scaffolding. Make it feel like a real product, not a project.

Building With AI (What Actually Happened)

I used ChatGPT and Codex throughout this process. But not in the way people often assume. There’s a narrative right now that you can just ask for an app and it appears. That wasn’t my experience.

The tools were powerful. They accelerated things. They handled details I would have struggled with on my own.

But they didn’t:

• Hold the product vision
• Make design decisions
• Understand tradeoffs in real contexts
• Decide what not to build

At best, they were collaborators. Which meant I still had to show up with clarity.

It’s not just about writing code; it’s about owning the vision of the app. And maybe more importantly:

This whole idea that ChatGPT can write an app for you—it’s not automatic.

You can move faster now. But you still have to steer.

The Workflow (If You Want to Try This Yourself)

If you’re curious about building your own app using a similar approach, here’s what actually worked for me:

1. Start with a real problem Not an idea. A friction you’ve felt repeatedly.
2. Define constraints early What will this not become? Hold that line.
3. Work in small, named sprints Each sprint should have a clear intention. Not just a feature, but a reason.
4. Build, test, reflect, repeat Don’t try to architect everything upfront. Let the product reveal itself.
5. Use AI as a collaborator, not a shortcut Ask specific questions. Iterate. Stay involved.
6. Protect the core experience Every new feature should earn its place without breaking what made the app work in the first place.
7. Ship when it feels coherent—not perfect There were plenty of things I could keep refining. At some point, you have to release it into the world.

Where This Landed

NextUp Music is now live!! It’s a small app. Intentionally so.

It doesn’t try to recommend music. It doesn’t try to optimize your listening. It doesn’t try to become a platform. It just creates a system where shared listening feels fair, personal, and calm.

And for me, that’s enough.

👉 App Store: https://apps.apple.com/us/app/nextup-music-player/id6757323472

👉 Stoked Logic: https://stokedlogic.com/#apps

If you try it, I’d be curious what you notice. Not just whether it works—but whether it changes how the moment feels.

Three weeks before my Bluehost renewal, I found myself staring at a question that had been quietly growing for years. Do I really want to keep paying for this?

Not because Bluehost had done anything particularly wrong. In fact, it had faithfully hosted my websites for almost exactly 20 years. But over the last few months, my workflow had changed dramatically. My Mac mini at work had become essentially a server for experiments and projects (and also just a daily computer). Cloudflare had just become my DNS provider after I moved Stoked Logic over to experiment with Cloudflare Tunnel. WordPress websites (all 17 WordPress websites!) had become less central to my grad school teaching as Canvas matured into the LMS I now use for adjunct teaching.

Meanwhile, Bluehost had become… expensive. The renewal notice wasn’t really the problem. It was the catalyst. It forced me to look at 20 years of accumulated web infrastructure and ask a harder question:

If I were building this today, would I build it this way?

The answer was an immediate but timid “no.”

The Architecture I Actually Wanted

Once I wrapped my head around how hosting no longer has to be “where my website lives,” the project became clearer. I realized I wanted each service to do one job well:

• Cloudflare for domain registration, DNS, SSL, proxying, caching, security, and Cloudflare Tunnel. (routing web traffic)
• KnownHost for reliable, boring WordPress hosting with cPanel. (hosting website stuff)
• Mac Mini for bots, experiments, development, and anything that didn’t belong on shared hosting. (internal scripts, tiny weblinks, network protocols, Claude Cowork tasks, etc)
• iCloud Mail for the domains where I actually receive email. (This is part of an iCloud+ subscription!)
• Simple forwarders for domains that exist mostly as aliases. (Who here still has an EduAwesome.com email address from me??)

Instead of one company doing everything adequately, I wanted each layer to do exactly what it was best at.

That sounds obvious now. It certainly wasn’t obvious when I started.

The Unexpected Role AI Played

People often ask whether AI writes my code. That’s actually the less interesting story. The interesting part was confidence.

Over the past several months I’ve built a very focused ChatGPT workspace around my infrastructure and development projects. By the time I started thinking about this migration, it already understood my domains, hosting providers, Git workflow, Xcode projects, Mac mini, Cloudflare setup, and long-term goals.That meant I wasn’t starting every conversation from scratch.

Instead, it felt like sitting down with an experienced systems administrator who already knew my environment. More importantly, it let me approach the project differently. Instead of feeling like I needed to understand DNS, hosting, registrar transfers, SSL, email routing, Cloudflare, and cPanel before making a single change, I only needed to understand the next step.

DNS first. Hosting second. Registrar transfers third. Email cleanup fourth.

Every step was verified before moving on. Every change was reversible. That changed everything.

Step 1: Move DNS to Cloudflare

This was actually the easiest part. Scariest, but easiest.

Cloudflare imported my existing DNS records, after which I spent some time cleaning them up (aka pasting DNS records from Cloudfare and from Bluehost into ChatGPT and getting advice on what to keep, change, verify).

Along the way I learned something that probably should have been obvious years ago: DNS, hosting, and domain registration are three completely different things. And while having all those three together in Bluehost was so much easier back in 2006, it left me feeling stuck at the bottom of an immovable mountain whenever I considered any chanes.

Put simply, I had mentally grouped all of them together because Bluehost happened to provide all three. Separating those responsibilities made the rest of the migration much easier, simpler, and safer.

I downloaded all 44GB and pointed Codex at them so it could actually evaluate and inventory every file, and every WordPress database (saved as MySQL). The cleanup included (with ChatGPT telling me what to ask Codex to review):

• removing years of obsolete cPanel records
• enabling modern SSL/TLS settings
• configuring DMARC
• simplifying SPF records
• proxying websites through Cloudflare
• leaving mail-related records as DNS-only

Once most of the stuff was migrated, ChatGPT suggested I run SSL Labs to get a rating of how secure each domain is. A particularly satisfying moment came when SSL Labs went from a B rating to an A simply by disabling TLS 1.0 and 1.1 and tightening a few Cloudflare settings.

Sometimes better security really is just a checkbox.

Step 2: Move Hosting to KnownHost

This part looked scary.It ended up being surprisingly straightforward.

Before asking KnownHost to migrate anything, I downloaded everything from Bluehost: all the websites and databases and files.

Having a complete offline backup changed the emotional equation. Once everything existed safely on my own machine, I no longer felt like I was operating without a safety net. A migration this big was new to me, so having a backup was a big deal. Pro tip: if you’re consider migrating to a more modern and cheaper host, definitely do this step.

But came an unexpected challenge. Of course. I discovered I had seventeen MySQL databases: some belonged to active WordPress sites, some belonged to abandoned experiments/websites, some were backups (I think), and some I honestly couldn’t identify anymore.

Rather than guessing, I worked through each one with Codex until we mapped every database back to its corresponding WordPress installation. (Pro tip: codex command was “Look at all 17 MySQL files and tell me which ones are active.” It was seriously that easy.)

By migration day (from Bluehost to KnownHost), the picture had become wonderfully simple. Three active WordPress sites. Four active domains (billselak.com, stokedlogic.com, eduawesome.com, headsdowntwothumbsup.com). Everything else could be archived.

Step 3: Transfer the Domains

Changing DNS is not the same thing as transferring a domain. Because Cloudflare was already serving DNS, the registrar transfer became almost boring. The websites stayed online. DNS never changed (cuz it already had a few days ago, and took like 5 minutes).

Only the company responsible for annual renewals changed. Today, Stoked Logic has already completed its transfer to Cloudflare Registrar. The remaining domains are simply waiting through Bluehost’s mandatory transfer window.

Things That Broke

Of course things broke. That’s part of the fun. 🫠

One of the more memorable moments came when stokedlogic.com displayed an Cloudflare Error 1000 after migrating hosting. It turned out the migration had copied the files correctly, but the domain itself hadn’t been created inside cPanel. Sounds scary, but again, I just had to tell Codex “Hey, getting Cloudfare Error 1000. Diagose and fix.” And it fixed most of it by itself and told me a couple things to do inside a couple files.

Once that was fixed, another issue appeared. WordPress was still convinced its home URL was Bluehost’s temporary migration address. Again, pasted screenshot of weird URL, and Codex told me how to resolve it. Fix that… and suddenly login cookies stopped working. Fix that… and everything came alive. Yay!

None of those individual problems were especially difficult with Codex. Having ChatGPT act as a troubleshooting partner meant each problem became a short debugging session instead of an evening spent wandering through forum posts from 2014.

The Part I Didn’t Expect

The hosting migration uncovered a much larger project. Buried inside those old WordPress installations were graduate courses I’d built between roughly 2008 and 2015. Looking back, they feel surprisingly modern.

At the time, Canvas wasn’t part of a teaching workflow. Google Classroom didn’t exist. There wasn’t a polished LMS that handled weekly agendas, assignments, resources, discussions, and course navigation the way we expect today.

So I built my own. Weekly agendas became WordPress Pages. Assignments became Posts. Resources lived alongside the lessons. The entire course unfolded publicly on the web.

At the time, this wasn’t simply following the crowd. It was part of exploring what online learning could become before today’s learning management systems had really matured (been invented?).

Today, I still teach graduate courses. The difference is that Canvas now handles the operational side of teaching beautifully.

But the ideas inside those older courses? Many of them are still worth preserving. Not because the screenshots are current.Not because the software recommendations still apply. Because good questions about teaching rarely expire.

Enter ArchiveSmith

Originally, I planned to manually copy everything into new WordPress pages. That idea lasted about ten minutes. I quickly realized I wasn’t migrating one website. I was trying to rescue nearly a dozen aging WordPress installs, each with its own pages, posts, media, redirects, and quirks.

So, naturally, I built a tool. Well… we built a tool. Working with Codex, I created a collection of scripts that eventually became ArchiveSmith.

It now:

• reads a WordPress SQL export (a WordPress site is a collection of page snippets and a SQL database)
• reconstructs pages and posts
• inventories linked media
• proposes redirects
• generates review documents
• creates WordPress import files
• prepares clean archival drafts

The first pilot converted an entire graduate course into a structured archive in minutes instead of days. Even better, it standardized the process. Every old site now follows the same review workflow before anything is published.

I started building it to solve my own problem. Now I’m wondering whether other educators, universities, museums, and nonprofits have exactly the same problem sitting quietly on an old shared hosting account somewhere. If so, ArchiveSmith might eventually become a small macOS app that helps rescue those sites before they’re simply deleted.

Looking Back

When this project began, I thought I was changing web hosts. Instead, I ended up simplifying 20 years of infrastructure, cleaning up DNS, modernizing security, preserving years of teaching materials, and building a new archival workflow I’ll use for every future migration.

The technical pieces are satisfying… Cloudflare. KnownHost. Cleaner DNS. Lower annual costs. Better security.

Those are all worthwhile outcomes. But they’re not what I’ll remember. Well, actually they will. But a better blog post ending is that what I’ll remember is realizing that AI didn’t replace understanding: it changed how I approached understanding. I didn’t need to know everything before I started. I needed enough confidence to take the next well-scoped step, verify it, and keep moving.

Sometimes that’s all the momentum a complicated project needs. And sometimes, somewhere in the middle of cleaning up 20-year-old DNS records, you accidentally build the first version of a tool that might help someone else preserve a little piece of the web.