Judgment, an AI infrastructure startup focused on improving AI agent performance through production data analysis, announced $32 million in Seed and Series A funding. The rounds were led by Lightspeed Venture Partners, with participation from Nova, Valor, Dynamic, Chris Manning, and the founders of DoorDash and Mercor.
Turning Production Data Into AI Improvements
As AI agents increasingly handle tasks tied to economic value, large volumes of operational data are being generated through reasoning tokens, tool usage, retries, memory, and user interactions. This production data provides direct insight into how AI agents behave across customer interactions, software environments, and enterprise workflows.
Judgment is developing infrastructure that transforms this operational data into actionable intelligence for AI teams, helping organizations identify recurring failures, monitor agent behavior, and improve system performance at scale.
Infrastructure For Monitoring And Optimization
Current AI development workflows often require teams to manually review logs, analyze traces, identify repeated failures, and map issues back to prompts, tools, retrieval systems, or agent logic. Judgment aims to automate that process through AI-powered analysis and evaluation infrastructure.
The platform enables AI agents to search, analyze, and interpret production data, generating evaluation systems, alerts, and continuous improvement loops designed to optimize real-world agent performance.
Expanding AI Agent Operations
The funding will support further development of Judgment’s infrastructure platform as demand grows for scalable systems capable of managing, monitoring, and improving AI agents operating in production environments.