LMS reboot
Long ago ... I hand-coded my own LMS similar to Moodle and installed it in a handful of schools. Life moved on and I worked up to Principal.
Now, retired, I'm fascinated by the power of AI coding agents, I'm using Claude Code, to redefine SaaS development. Just 12 months ago, lack of coding expertise was a terminal barrier. Ai has potentially eviscerated that.
Can I use AI (Claude Code) to build an industry standard LMS, with domain expertise, but no coding support to build a (near) zero-cost, fully featured competitor to the likes of Canvas, Schoolology, etc?
In 2 weeks, I had v1 working. A bespoke solution for a single client. In 4 weeks, I had v2 working, a version for multiple clients. Now, I'm working on v3 - a customiseable product that can be skinned and styled to meet a variety of education briefs with its own SCORM authoring tools.
It runs on Node.js with SQLite, deployed on Railway, and includes SCORM integration, student progress tracking, PDF workbook submission, digital submission using online markup tools, AI grading, customisable chatbot customised to client's knowledge base, online SCORM authoring tools, quizzes.
I built it entirely with Claude Code. Every feature, every database migration, every deployment issue — worked through in conversation Clause.
Some things that have surprised me:
Speed: Features that would have taken a professional team weeks, I've shipped in days.
Debugging: Claude Code reasons through problems. DNS issues, mobile compatibility bugs, database persistence — we've worked through all of it.
Cost: I've required Claude to use free or very cheap development options and have, so far, achieved this.
No ceiling: I have yet to have a feature that "I" can't build. It's insane.
Schoolhouse is far from taking on Canvas, but the speed and grace of development makes me curious whether AI tools have shifted the economics of SaaS enough that a domain expert working solo can build something viable.
I'm looking for early testers to kick the tyres and give feedback.
