Case studies

Real engagements, fully anonymized. What an audit or rescue of an AI-built or no-code MVP actually looks like in practice — scope, timeline, what changed, what we'd tell a founder in the same situation to watch out for.

// Rescue·Lovable → AWS·2–3 weeks

From Lovable to production in 2 weeks

A pre-launch PropTech SaaS, built end-to-end in Lovable, with no production infrastructure, no CI/CD, no monitoring, and no security review. Two to three weeks later: OWASP-audited, automated deploys, 24/7 monitoring, file storage migrated to AWS S3 + CloudFront, handoff docs for the client's team.

Lovable → AWSPre-launch2–3 weeks

Read the case study →

More case studies coming as engagements complete. Every case study is published anonymized by default — no client names, no identifying details. If you're curious about a specific stack or situation that isn't listed here yet, ask on the discovery call and I'll tell you whether I've seen it.

Not a case study, but worth reading: my articles on what goes wrong in AI-generated code, the hidden costs of a "cheap" MVP, and when to hand off your MVP — the patterns case studies like the one above show up in, explained in general.