Why AI-generated applications fail in production
Preview builds hide the failures that show up once real users, real data, and a real deploy pipeline are involved.
Blog
Technical writing
Engineering notes from the work: why generated apps fail after preview, and the checks we run before a launch.
Common Supabase RLS mistakes in AI-built apps
The usual Supabase security failures are disabled RLS, service-role keys in the client, and policies that trust a user-controlled column.
How to secure a Next.js application before launch
A practical Next.js application audit: server actions, secrets, auth boundaries, and what not to put in the client bundle.
Lovable deployment troubleshooting
What to check when a Lovable app works in preview and fails on the deployed URL.
How to prepare an AI-generated app for launch
A short launch sequence for apps built with Bolt.new, Lovable, Replit, Cursor, or Claude Code.
Production-readiness checklist for AI-built apps
A checklist for Next.js, Supabase, and Vercel apps before the first real customer.