Hi HN, I’m building Codela to help coding agents learn from what happens after their code ships.

I realized that my AI agent writes a change, the tests pass, and it gets deployed. Then in production a problem happens, I fix it, but what it did wrong and why never reaches the agent, so futures session usually make the same mistake.

Codela checks every deploy against how your app actually behaves in production, learns from the ones that broke it, and teaches your agents so they don’t repeat the mistake.

Would love any feedback on this, thanks!