Phoenix

Phoenix is an open-source platform designed for AI engineering teams to develop, trace, evaluate, and iterate on autonomous AI agents. Supporting full observability, experiment tracking, error annotation, and evaluation workflows, Phoenix enables systematic quality improvement for AI agents in production. It caters primarily to AI engineers, MLOps teams, and enterprise organizations looking to confidently ship and maintain LLM-based or agentic AI systems in a privacy-respecting, vendor-agnostic, and standards-based environment.