Observability

Langfuse

Langfuse for tracing, prompt management, evals, cost and latency, and debugging.

When we use it

  • You need traces across multi-step agent and tool runs.

  • Prompts should be versioned and managed outside the code.

  • You're running evals and tracking cost and latency.

  • Production behavior needs to be debuggable after the fact.

Newsletter

One letter, every week. Working systems — not hot takes.

Build logs, agentic engineering decisions, agent failures, evals, and what survives real users. Sent weekly, never more.

Weekly. No spam. Unsubscribe anytime.