Courselet

The problem

One of our clients was paying for a third-party service to embed a micro-course inside an app we had built for them. A course builder of our own could replace it down the line: a real product with a real reason to exist, but no client waiting on it.

That made it the right first test for our AI development flow. A flow on paper is worth nothing until it survives a real project, and we did not want its first run to be on a client.

What I built

An LMS is heavy and a slide deck just sits there. Courselet is for what sits between: a coach’s method, an onboarding explainer, a short briefing, finished in one sitting on a phone.

• Split-screen editor

  Creators build lesson and quiz cards side by side with a preview that shows exactly what a learner will see.

• Immutable publishing

  Publishing freezes a snapshot of the course to a clean public link, so later edits never change what learners are halfway through.

• No-account learners

  Learners tap through, answer quizzes with instant feedback and hit a completion screen. No account, no app, no personal data collected.

• Brand as tokens

  The design is a token system down to the signature progress bar, where each pill is literally one card of the course.

My role

Lead developer, about 94% of commits, essentially the whole MVP. My day was not watching code stream past: I was managing agents, making sure none was sitting on a blocker, testing what they put out, and working on a separate project at the same time.

The flow started with documents (requirements, brand, journeys, page specs and a clickable prototype) drafted by AI, reviewed, then locked. Then the agents built in parallel, everything through pull requests.

Key decisions

1. Documents before code

   If something was not written down, agents flagged the gap instead of guessing. Locking the docs early is what made parallel building safe.

2. Snapshots, not live courses

   Publishing writes an immutable snapshot, which keeps the learner side simple, cacheable and safe from half-finished edits.

3. Collect nothing from learners

   Anonymous by design. No sign-up means no friction for learners and no personal data to protect.

Outcome

Shipped

A working, scalable app, live at courselet.oneeleven.tech.

Flow proven

Two weeks later the same flow ran a real client project.

One honest gap

An internal run has no client UAT. The next run did.

Stack