Observability for AI applications in production

An AI app in prod has 4 failure modes classic monitoring doesn't see: variable latency, token cost drift, hallucinations, model drift. Learn to watch them with Sentry, PostHog, Snyk Agent Scan, and the right architecture patterns (rate limit, cache, fallback).

Level : intermediate · Duration : 360 min

Modules

1. 1. Why observing an AI app is different
2. 2. Sentry, front + back errors, source maps and Session Replay
3. 3. PostHog, product metrics, AI funnels and feature flags
4. 4. Agentic security, Snyk Agent Scan + CodeBurn
5. 5. LLM architecture patterns in prod, continuous eval, fallback, caching

View on Coeurdar