Judgment Labs

Infrastructure for monitoring and improving the behavior of AI agents in production. Judgment Labs positions itself as a continuous-improvement layer: it captures long agent trajectories — tool calls, retries, memory updates, missed clarifying questions — detects recurring failure modes across that production data, turns them into evaluations and alerts, and feeds fixes back into the agent. The platform spans tracing, behavior discovery, trajectory-level "judge" evaluators, and regression alerting, and includes an open-source framework, Judgeval. It targets teams building deep agents in domains such as legal, enterprise support, and financial AI. Founded in 2025 by three childhood friends: CEO Alex Shan (previously an AI researcher in Stanford's NLP group under Christopher Manning), Chief Scientist Andrew Li (an early research hire at Together AI), and CTO Joseph Camyre (formerly an infrastructure engineer at Datadog). The roughly 22-person team is based in San Francisco and counts E3 Group, Monaco, Human Behavior, and Contrario among its users. Judgment Labs raised $32 million across combined seed and Series A rounds in May 2026, both led by Lightspeed Venture Partners, with Nova Global, SV Angel, Valor Equity Partners, and Dynamic participating. Legal entity: Judgment, Inc.