Break Nothing: How to Ship & Evaluate AI Features Safely

​Shipping AI features is risky, small changes in models or prompts can break user experiences in unexpected ways. In this joint session, Rogerio Chaves (CTO, LangWatch) and Asaph Kotzin (Principal PM, Flagsmith) will show how combining feature flags with AI evaluations lets teams safely test, release, and monitor AI-driven functionality.

​Learn how to roll out AI updates with control, catch regressions before users do, and build reliable release workflows for prompts, models, and agents. Perfect for teams bringing AI into production and looking to ship faster—without breaking anything.