Judgment Labs (@JudgmentLabs) / X

Judgment Labs

106 posts

Judgment Labs

@JudgmentLabs

Improve your agents from production data.

Joined April 2025

Pinned
Judgment Labs
@JudgmentLabs
May 12
Today is a special day.
Alex Shan
@alexshander03
May 12
We’re launching @JudgmentLabs today and announcing $32M in funding. As AI agents take on more of the work that creates economic value, they generate massive amounts of production data: the clearest record of how they behave with users, software, and the real world. Judgment
00:00
2.3M
Judgment Labs
@JudgmentLabs
Jun 5
Bay to Breakers is the city's biggest footrace... And Judgment took it over. 🟠 30 of us. 7.46 miles. This is the Summer of Judgment. Text "JL" to (628) 888-7594 to get on the list for our next event.
00:00
48K
Judgment Labs reposted
Enyu
@0xhappier
Jun 5
We ran Bay to Breakers. 30 of us. 7.46 miles. One Sunday morning in San Francisco.
30K
Judgment Labs reposted
Shun
@22uenos
May 28
Had a lot of fun making these visuals :) Made in @paper with Claude
Judgment Labs
@JudgmentLabs
May 28
We built Agent Judge to evaluate long-horizon agents. As agents take on longer tasks, the evidence needed to evaluate them gets buried across tool calls, retries, logs, database updates, and final outputs. Evaluating these agents requires investigating the trajectory, not just
66K
Judgment Labs
@JudgmentLabs
May 28
Replying to @JudgmentLabs
The core idea: long-horizon agent evals should be done by agents, not simple LLM judges. Agent Judge searches trajectories, verifies stateful actions, and adapts rubrics from production feedback.
Agent Judge: Solving Long-Context Evals for Production Agents
From judgmentlabs.ai
4.8K
Judgment Labs
@JudgmentLabs
May 28
Replying to @JudgmentLabs
Agent behavior changes as models, tools, products, and user workflows change. That means the rubric used by the judge has to improve from production data, so it keeps evaluating the behaviors that matter. Rubric Builder turns feedback into concrete rubric updates.
700
Judgment Labs
@JudgmentLabs
May 28
Replying to @JudgmentLabs
Production agents also change state. If an agent claims it updated a CRM, opened a PR, changed cloud config, or triggered a workflow, the eval should verify what actually happened. Agent Judge can inspect tool evidence, database logs, audit trails, event streams, and other
503
Judgment Labs
@JudgmentLabs
May 28
Replying to @JudgmentLabs
Simple LLM judges break because long-horizon trajectories do not fit into a context window. They either see a narrow slice of the run, or try to ingest a long dense trajectory and miss the evidence in the middle. Agent Judge gives the evaluator the ability to search long
644
Judgment Labs
@JudgmentLabs
May 28
We built Agent Judge to evaluate long-horizon agents. As agents take on longer tasks, the evidence needed to evaluate them gets buried across tool calls, retries, logs, database updates, and final outputs. Evaluating these agents requires investigating the trajectory, not just
74K
Judgment Labs
@JudgmentLabs
May 27
Thrilled to be recognized by @Redpoint as one of the most promising private AI Infrastructure companies. More exciting news to come!
Redpoint
@Redpoint
May 27
The Redpoint InfraRed 100 is now live. These are the companies building the infrastructure that powers everything happening in AI right now, from world models and agent runtimes to the sandboxes, databases, and security tools agents depend on. Congratulations to this year's
2.8K
Judgment Labs reposted
Judgment Labs
@JudgmentLabs
May 18
@tbpn thanks for having our team on! P.S. it's Judgment 🧡
00:00
256K
Judgment Labs
@JudgmentLabs
May 16
Scoops will start flying @ 1pm (500 Marina Blvd) Tomorrow @Baytobreakers Finish Line The Flavors: - Fudgement @JudgmentLabs - Berry Brex-fast @brexHQ - Modal Green Tea @modal - Claude au Lait @claudeai - Vercel Road @vercel - Ando Apple Pie @andocorporation
Judgment Labs
@JudgmentLabs
May 16
Summer of Judgment! 2 days left for free ice cream...
00:00
2.7K
Judgment Labs
@JudgmentLabs
May 16
Summer of Judgment! 2 days left for free ice cream...
00:00
315K