Zero-code full-stack observability for AI agents and LLM apps. OpenTelemetry-native monitoring for LLMs, VectorDBs, and GPUs with built-in guardrails, evaluations, prompt hub, and a secure vault. Fully self-hostable anywhere.