From private sessions to a movement platform

14 sessions in, and the thing we built for one person is starting to look like a platform.

I wrote a few months ago about the technical architecture behind this project — how we extract coaching sessions into structured move documentation using a three-layer AI system. That post was about the how. This one is about the what-now.

Because somewhere between session 10 and session 14, we crossed a line. We stopped building a personal reference library and started building something other people could actually use.

The numbers tell the story

84 moves across 14 sessions. That's not a handful of stretches — it's a curriculum. My coach Jon has been teaching movement at standing desks for years, and the knowledge base we've built from our sessions is starting to represent a real chunk of his methodology.

Each move has setup instructions, progressions, dosage guidance, and intention profiles at four levels. It's not a video library where you watch someone do a thing. It's structured coaching data that adapts to how much attention you have right now.

That distinction matters, because it's the basis of the business model we're figuring out.

Three tiers, one flywheel

Here's the model we're working toward. Three tiers, each one adding more personalization:

The interesting part isn't the tiers themselves — tiered pricing is as old as SaaS. The interesting part is what connects them.

The Q&A flywheel

Here's the thing that got me excited enough to write this post.

Every tier generates questions. "My shoulder clicks during this move." "Can I do this sitting?" "What's the difference between the isometric and the eccentric version?" Normal coaching questions.

In a traditional model, Jon answers every question manually. That's fine for 10 clients. It's impossible for 500. And it means his knowledge stays locked in one-off replies that nobody else benefits from.

What we're building instead:

1. Subscriber asks a question
2. AI drafts a response from the knowledge base — 84 moves of structured coaching data, plus Jon's teaching patterns from 14 sessions of transcripts
3. Jon spot-checks the draft. Approves, tweaks, or rewrites
4. Response goes to the subscriber
5. The interaction — question, answer, and any corrections Jon made — feeds back into the knowledge base
6. Next similar question gets a better draft. Jon reviews faster. Eventually, some categories barely need review at all

Every interaction makes the system smarter. Not in a hand-wavy "AI learns" way — in a concrete, measurable way. The correction log grows. The extraction prompts improve. The knowledge base fills gaps we didn't know existed because nobody had asked about them yet.

Jon's time shifts from answering questions to quality-checking answers. His expertise scales without him having to repeat himself.

Why this isn't a course

I've bought online courses. You probably have too. Most of them are static — recorded once, sold forever, slowly going stale. The instructor moves on. The content doesn't evolve.

This is different because the platform gets better the more people use it. A question about knee pain during a squat pattern doesn't just help the person who asked — it makes the system better at handling knee pain questions forever. Jon's coaching intelligence compounds.

It's also different because Jon and I are still doing sessions. New moves get added. Existing moves get refined. The knowledge base isn't a snapshot of what Jon knew in April 2026 — it's a living document that grows every week.

The honest part

We don't know if this works yet.

We don't know if people will pay $25 a month for standing desk workouts. We don't know if the AI Q&A flywheel will actually reduce Jon's review time fast enough to matter. We don't know if 84 moves is enough to cover most people's needs or if we need 200.

What we do know: the system works for me. I haven't missed a workout in years. I train at my desk while I work. My body feels better at 40-something than it did at 30. And the documentation system we built captures Jon's coaching well enough that I can reconstruct a full session from the data alone.

That's a sample size of one. Not exactly a peer-reviewed study.

But it's enough to keep building. And building in public means you get to watch us figure out the parts we haven't figured out yet — pricing, onboarding, what happens when someone with a very different body tries moves designed around my proportions.

We'll share what works. We'll share what doesn't. That's the deal.