I’m curious what others are actually running in production today—custom frameworks, LangGraph, AutoGen, CrewAI, homegrown pipelines, etc. and how satisfied you are with them.
What’s working well? What breaks down at scale (debuggability, cost, determinism, tool orchestration, evals)?
sshadmand•23h ago