today, we're introducing self diagnostics: the first ever way for agents to proactively self-report issues they encounter.
welcome to the future of agent observability.
00:00
- today, we're introducing self diagnostics: the first ever way for agents to proactively self-report issues they encounter. welcome to the future of agent observability.00:00
for years, agent observability meant spending weeks making eval sets that break every time a new model comes out. ~1 year ago, @raindrop_ai introduced a better way - tiny classifiers that detect anomalies in user/agent interactions. but a lot has changed since then.agents have gotten a lot smarter. they often understand themselves better than you do. critically, they also have access to debugging context that is not even available post-facto (e.g. reasoning traces are encrypted by OpenAI/Anthropic) the best place to detect and diagnose
