For those of you running agents in production, I'm trying to understand how teams catch issues like prompt regressions, latency spikes after model swaps, or edge cases that only show up on real calls. Put together a quick survey to map what people actually do (vs what we all know we should do 😆 ). 3 min, results will be shared with participants: https://docs.google.com/forms/d/e/1FAIpQLSe20-dyoMIRgnrDxkiAxc5bCBcL0S5LLYy6PGNb3HcB6N2tTg/viewform?usp=sharing Thank you all!