Judgment Labs

Judgment Labs builds Judgment, a monitoring and improvement platform for long-horizon AI agents. It stores and indexes agent traces, surfaces recurring failure patterns, supports hybrid search across traces, and lets teams test fixes with offline evals. Tracing covers subagents, tool calls and spans, and monitoring runs filters, code checks, decision models and agentic judges. The product is also available through MCP and a Slack agent.

Customers are teams building AI agents; the website names Monaco, Listen Labs and Strala among users. Pricing is usage-based: a free Beginner tier, a Pro plan at $250 per month with pay-as-you-go usage beyond its credits, and custom Enterprise plans. Usage is billed for trace ingest per GB and for evaluator runs per 1,000.

Enterprise deployments can run inside the customer's own AWS or GCP account. The company states SOC 2 Type II and HIPAA compliance. It announced $32M in funding led by Greenoaks and Lightspeed in May 2026. Its open roles are based in San Francisco, covering agent engineering, deployment engineering and business development.

Product: Judgment: agent trace monitoring, issue discovery, hybrid search, offline evals and judges for long-horizon AI agents.