Judgment Labs just opened its doors to everyone. The San Francisco startup (official website), which builds monitoring tools for AI agents, has moved from a closed pilot to general access, so any team running agents can now use it to catch failures before customers do.
The pitch is simple. AI agents are getting longer-running and more complex, and the old ways of checking them are falling behind. Judgment Labs wants to be the layer that watches every step an agent takes and tells developers when something breaks.
Here is the launch announcement as it appeared on X:
Judgment Labs launch announcement on X
Key Points
- Judgment Labs now offers general access to its AI agent monitoring platform.
- The tool watches agent behavior and flags failures before users notice them.
- The startup raised $32 million in Seed and Series A funding, led by Greenoaks and Lightspeed.
- Alex Shan, Andrew Li, and Joseph Camyre founded the company.
What Judgment Labs Actually Does
Judgment Labs calls its approach agent behavior monitoring. Every time an agent runs, it leaves a trail: each model call, each tool it touched, what it was asked, what it answered, how long it took, and what went wrong. Engineers call that trail a trace.
The platform collects those traces, scores them, and flags the runs that look wrong. Teams see how often a problem shows up in live traffic and can open the exact traces behind it. The system also groups similar failures on its own, which helps teams spot patterns they never thought to look for.
Why Agent Data Is a Big Deal
The company argues that long-running agents will create more exhaust data than anyone can process today. Hidden in that data is a clear record of what users want, where agents stumble, and how they behave when they meet the real world.
That is the opportunity Judgment Labs is chasing. Instead of treating traces as logs nobody reads, it treats them as the raw material for making agents better.
The Problem Early Customers Faced
Before launch, Judgment Labs spent months working side by side with a small group of companies that run advanced agents. Those agents changed constantly. Their testing and monitoring setups could not keep pace.
Reviews happened by hand. Worse, customers often reported the failures first. Judgment Labs says its platform now flips that order: it detects agent issues early and flags them for a fix before users notice.
Key Features of the Platform
- Full tracing: Records every decision, tool call, and result in an agent run.
- Production monitoring: Checks live traffic for known bad behaviors and alerts teams when an agent regresses.
- Issue discovery: Surfaces new failure patterns that nobody defined in advance.
- Judgment Agent: A built-in assistant that searches traces and helps find the root cause.
- Offline tests: Turns real failures into tests so the same bug stays fixed.
- MCP server: Lets coding agents reach Judgment project data and tools.
Developers can start with the official quickstart guide. The company also runs an open-source toolkit called Judgeval, which lives on GitHub.
Meet the Founders Behind Judgment Labs
Three co-founders run the company. Alex Shan is the CEO, Andrew Li is the Chief Scientist, and Joseph Camyre is the CTO.
Shan studied at Stanford and worked as a researcher at the Stanford Artificial Intelligence Laboratory before starting the company. He is still in his early twenties, which makes Judgment Labs one of the youngest-led startups in the agent tooling race. You can follow him on LinkedIn and X.
Funding: $32 Million From Top Investors
Judgment Labs stepped out of stealth in May with $32 million in combined Seed and Series A funding. Greenoaks and Lightspeed led the round. The Lightspeed company page describes the startup as a builder of the continuous-improvement stack for AI agents.
Why This Launch Matters for AI Teams
More businesses now put AI agents in front of real customers. When an agent fails quietly, it can give wrong answers, burn money, or hurt trust, sometimes for days before anyone notices. Monitoring is turning into a must-have part of the AI stack, not an extra.
With general access live, small startups and big companies can now use the same approach the early partners tested. Judgment Labs says it is excited to see what builders create, and its message to them is short: give your agents better judgment.
Frequently Asked Questions
What is Judgment Labs?
Judgment Labs is an AI startup that builds a platform to monitor AI agents, find their failures in live use, and help teams fix them.
Is Judgment Labs open to everyone?
Yes. The company announced general access after working with a select group of early customers.
How much has Judgment Labs raised?
It raised $32 million in Seed and Series A funding, led by Greenoaks and Lightspeed.
Who founded Judgment Labs?
Alex Shan, Andrew Li, and Joseph Camyre founded it. Shan is the CEO.
