Opik

Opik logs every step an AI agent takes, from user interactions to context retrieval and tool calls, and adds automated evaluation workflows to find and fix errors in development, testing and production.

It provides agent tracing and debugging, LLM-as-a-judge evaluation, prompt management and production monitoring with dashboards. Evaluations can be run on datasets during development and on live traces in production.

Opik is developed by Comet. It has an open source version under the Apache 2.0 licence that can be self-hosted, a free tier on the hosted service and an enterprise platform.