Back to case studies
2026Live4 weeks · Mar–Apr 2026

CourseAI

AI-generated structured courses from YouTube, PDF or raw text — shipped to paying subscribers in 4 weeks.

Role
Full-cycle SaaS · spec to paid subscribers
Open project ↗

By the numbers

4 weeks
Spec to paid subscribers
Stripe
Subscription billing live
3
Input formats (YT / PDF / text)
ADR-0001
Auth pivot documented

Stack

  • Next.js 15
  • React 18
  • TypeScript
  • Prisma
  • Neon Postgres
  • NextAuth
  • Stripe
  • Google Gemini
  • Tailwind
  • Vercel

Process receipts

The pipeline of agents that built it, the count of commits and deploys, and the wall-clock time from spec to production. No decks — just the audit trail.

4 weeks
Shipped in
180
Commits
12
Deploys
7
Agents in pipeline
Agent pipeline
  1. product-manager
  2. business-analyst
  3. architect
  4. prisma-expert
  5. developer
  6. reviewer
  7. test-engineer

Context

Creators had raw knowledge in YouTube transcripts, PDFs or long-form docs but no scaffolding for learners. We built a platform that ingests any of those inputs and emits a structured course: modules, lessons, quizzes, reading-time estimates. Shipped to paying subscribers in one month.

Approach

Next.js 15 App Router with server actions for ingestion, Gemini for generation, Neon for state. NextAuth for email-only auth after we hit a dead end with Google OAuth (documented as ADR-0001). Stripe for subscription billing. Tailwind for UI.

Stack decisions

Gemini over OpenAI for cost at the target quality bar — course generation is a long-context task where Gemini Pro wins on price-per-token. Neon over Supabase for pure Postgres + branching on PRs. Vercel for deploy to pair with Next 15 edge features.

Outcome

Live at the link above. Four weeks from empty repo to first paying subscriber. Stripe subscription billing, email-only auth, course generation from any of three input formats. What we ship for clients with similar SaaS briefs.

Have a similar project?

We reply within 24 hours. Usually faster.