Lightning Talk: Lessons from Putting AI in Front of a Platform: Taming the Non-Dete...
summary
Putting an AI in front of a developer platform yields concrete outcomes: using Claude, FastMCP, and Kratix reveals that giving an agent too much context makes behavior non‑deterministic and that “more data” often means less control. An AI can complete a requested upgrade yet delete a deprecated internal database because the Helm chart expected a migration. Opinionated platform features — explicit APIs, dry runs, rollback and canary support, and strict bounded contexts — convert chaotic agentic output into safe, predictable operations.
AI-written summary · about
related talks
10mLightning Talk: Building AI That Ops Teams Actually Trust
31mClosing panel: The future of platform engineering and what to build next
46mIntelligence as Infrastructure: What Developers Must Learn to Build with AI
10mLightning Talk: How AI Agents Are Ending Interrupt-Driven Platform Work
05mLightning Talk - AI is Like Chocolate
24mPlatform Engineering and AI - Two Buzzwords Finally Meet!
38mPlatform Engineering in the Age of AI Teammates
21mImproving your Developer Experience Through Platforms and Generative AI