Judgment is now publicly available.Get started

Announcing $32M in Funding Led by Greenoaks and Lightspeed

Infrastructure for improving AI agents from production data.

Judgment Labs Team2 min read

Today, we're announcing $32 million in funding to build infrastructure for improving AI agents from production data. Our Seed financing was led by Lightspeed Venture Partners and our Series A was co-led between Greenoaks and Lightspeed, joined by Nova, Valor, Dynamic, Chris Manning, as well as the founders of DoorDash and Mercor.

As AI agents take on more of the work that creates economic value, they generate an enormous amount of production data: reasoning tokens, tool calls, retries, memory, and more. This data forms the clearest record of how agents behave with customers, software, and their broader environments.

Production data is the highest-signal source for improving agents. When processed correctly, it reveals what users actually ask for, which failures repeat, and where agents succeed. But today, turning that data into improvements is still mostly manual. Teams read traces, scan logs, search for patterns, write evals, and map failures back to prompts, tools, retrieval systems, and agent logic.

Judgment builds infrastructure that enables agents to search, analyze, and understand your agents' production data. We help teams monitor agent behavior, uncover hidden failure patterns, and turn real-world usage into the evals, alerts, and improvement loops that make their agents better.

If you're excited about our mission, talk to us.