Case Studies
These are reconstructed engagements: invented teams, real dynamics. Every study follows the same arc — the vague prompt someone actually typed, what it produced or would have produced, the spec that replaced it, and at least one mistake that cost real hours. The mistakes are the point. A case study where everything worked teaches nothing you couldn't get from a playbook.
How to read them: steal the intervention points, not the stack choices. Whether a team used Fastify or Express is incidental. Where the human stopped the agent, what they checked before merging, and which claim they failed to verify — that transfers directly to your codebase. Each study ends with numbered lessons; if a lesson feels generic, we failed and you should tell us.
| Study | Scenario | Stack | Headline lesson |
|---|---|---|---|
| SaaS MVP | Two founders build a gym-equipment maintenance SaaS in 3 weeks | Next.js, Fastify, Postgres, Stripe | A 6-hour demo is not 20% of an MVP — it's 2%. Restart with specs, not momentum |
| Node Backend Refactor | Untangling a 45k-LOC, 5-year-old Express billing backend | Node, Express, MySQL, Jest | Pin current behavior — including the bugs — before letting an agent "improve" anything |
| Add Authentication | Retrofitting real auth onto a product with 30 paying customers | React, Fastify, Postgres | Auth is the worst feature to hand an agent vaguely; happy-path tests hide security holes |
| Admin Dashboard | 12 CRUD/ops screens for support and operations | React, TanStack, Fastify | Build one exemplar screen carefully, then mass-produce — and regenerate, don't patch |
| Prototype to Production | Hardening a deliberately dirty prototype before 200 users arrive | Express, BullMQ, Postgres | Verify cross-cutting claims per surface, not per PR — "added error handling" rarely means everywhere |
| CI/CD Pipeline | Standing up a full pipeline with agents writing the workflow code | GitHub Actions, Docker, Terraform | Pipeline code is code: it needs review gates just like the app it ships |
| Internal Platform | A platform team builds golden-path tooling with agent leverage | Go, Kubernetes, Backstage | Internal tools tolerate agent speed better than external products — but conventions must come first |
| Architecture Recovery | Reverse-engineering an undocumented system into decision records | Java, Spring, Oracle | Agents are excellent archaeologists if you make them cite file-level evidence for every claim |
| Multi-Agent Feature | Parallel subagents deliver one feature across API, UI, and tests | TypeScript monorepo, Claude Code subagents | Parallelism pays only when interfaces are frozen before agents fan out |
| AI-Built Product | A product where ~90% of code was agent-written, end to end | Next.js, Postgres, Claude Code in CI | The human's job shifts entirely to specs, review, and verification — and that job gets harder, not easier |