Latitude

Latitude is an open-source observability and monitoring platform for AI agents. It traces what your agents do in production, analyzes completed sessions to understand what happened, groups recurring failures into prioritized issues, alerts your team and can dispatch a coding agent to fix problems. The company describes the goal as making AI agents self-healing. It also offers an MCP server so AI tools can work with Latitude data directly.

Why agent observability matters

AI agents behave less predictably than traditional software. The same input can produce different outputs, tools can fail silently and conversations can go wrong in subtle ways, such as a user losing trust or abandoning a task. Standard logging rarely captures these problems. Latitude focuses on the behaviour of agents over full sessions rather than individual requests.

Conversation intelligence

Latitude analyzes completed sessions to extract what the conversation was about, what happened and where problems occurred. It detects moments like escalations, resolutions, abandonments, trust breaks, retries and tool failures, the situations that deserve a human's attention.

Signals and issue discovery

Failed traces become signals. Recurring failures are grouped and prioritized automatically, with trends and example traces attached, so teams see patterns instead of isolated errors. Alerts can be sent via Slack, email or webhooks when a new issue is detected or an existing one escalates, ideally before users complain.

Agent dispatch

Instead of only creating a ticket, Latitude can pass the problem with its context to your coding agent, which can propose or implement a fix. Engineers remain responsible for reviewing and merging changes.

Key features

• Production tracing for AI agents.
• Session analysis and behaviour pattern detection.
• Automatic grouping and prioritization of failures.
• Alerts via Slack, email and webhooks.
• Coding agent dispatch.
• Open-source code on GitHub and an MCP server.

Pricing

The Starter plan is free, with 20,000 credits per month, 30 days of data retention, unlimited seats and community Slack support. The Pro plan costs 99 US dollars per month with 100,000 credits per month and more features for teams building AI products at scale. Enterprise options are available, and the open-source version can be self-hosted.

Things to consider

Tracing production conversations means storing user data, so review retention settings and privacy obligations. Automated fixes should go through normal code review and testing.

Who it is for

Latitude suits engineering and AI teams running LLM-powered agents in production who need visibility into failures and a faster loop from detection to fix.