Judgment
Judgment ingests an AI agent's traces, applies evaluator judges to label failures, and offers an in-product agent that writes queries over traces, labels, datasets and test runs.
What it does
- Evaluation and observability
Features 5
- Long-horizon agent tracingTraces agents of any depth, including subagents, tool calls and spans, up to hundreds per trace.
- Hybrid trace searchSearch combines structured and semantic filters with customizable signals across agent traces.
- Custom signals and judgesSignals can be built from filters, code checks, decision models or agentic judges and run against traces.
- Offline evalsRecurring failures are surfaced and fixes can be tested offline before deployment.
- Bring-your-own-cloud deploymentJudgment can run inside the customer's own AWS or GCP account rather than only on the vendor's cloud.
Sources
- Pricing
-
Pricing page: "From your first experiment to your production deployment. Start free, pay for what you use."
- Description
-
Official documentation · 11 Nov 2026
Product page: "The Judgment Agent reads your traces, judge labels, Issues, datasets, and test runs. You ask in plain words and it writes the queries."
- What it does
-
Official documentation · 11 Nov 2026
Product page: "The Judgment Agent reads your traces, judge labels, Issues, datasets, and test runs. You ask in plain words and it writes the queries."
- Sold within
-
Sold on its own with published plans.
- Name
-
Official documentation · 11 Nov 2026
Name as published on the product page.
- URL
-
Official documentation · 11 Nov 2026
Product page, read 2026-11-11.
- Deployment
-
Official documentation · 11 Nov 2026
Product page: "SSO, SAML, and bring your own cloud"
- Interfaces
-
Official documentation · 11 Nov 2026
Product page: "SSO, SAML, and bring your own cloud"
Open Judgment Labs in the directory
Something wrong here? Send a correction — quote this product id: judgment-labs-judgment.