Latitude

What is Latitude?

Latitude is an open-source AI agent observability and monitoring platform. It traces agents running in production, finds failures, alerts the team and can send failure context to a coding agent to propose a fix. It is built for developers and teams building agents, and is MIT licensed with its code on GitHub.

Key capabilities of Latitude

• Conversation intelligence: analyzes sessions to surface escalations, abandonments, retries and tool failures.
• Issue discovery: groups similar failing traces into issues with trends and affected users.
• Alerts: notifies you when an issue appears or escalates.
• Custom signals: track specific behaviors such as failure modes or policy checks.
• Dataset management: versioned datasets from validated production traces for evals and regression tests.
• Human annotations: inline feedback on traces, spans and outputs.
• Agent dispatch: sends failure context to Claude Code or Cursor, which proposes a fix and opens a PR.

How Latitude works

You send traces with the Latitude SDK or an existing OpenTelemetry pipeline. Latitude analyzes sessions, clusters failures into issues and alerts you by Slack, email or webhook. From an issue you can build datasets for evaluations and hand the context to a coding agent that opens a pull request on GitHub.

Who uses Latitude?

Engineering teams running agents in production use it to see what breaks and why. The site lists SOC 2 Type II, SSO/SAML, GDPR tooling, EU data residency and audit logs for security-minded buyers.

Latitude pricing

Latitude is open source under the MIT license, so you can self-host it. Hosted plan names and prices are on the vendor's separate pricing page and are not published in the product overview.

Latitude alternatives

Hamming AI and Coval focus on testing and evaluating voice and conversational agents before and after launch. Domino Data Lab is a broader enterprise MLOps platform for building and governing models. Latitude is narrower, centered on production agent traces and failure discovery.