Latitude
by Latitude Data S.L.
Open-source observability and evaluation for AI agents — find and fix failures before production
Latitude is an open-source (MIT) observability and quality platform for AI agents that captures execution traces and turns raw logs into issue-style failure modes with evaluations attached. It helps AI engineering teams discover and fix why agents fail in production before users are affected.
Latitude is an open-source (MIT-licensed) observability and quality platform for AI agents, built by Latitude Data S.L., that helps engineering teams find and fix failure modes before they reach production. Rather than just storing logs, it captures full agent trajectories and clusters similar failing traces into actionable, issue-style failure modes with states and evaluations attached. Core capabilities include semantic search across 100% of traces without sampling, conversation intelligence that surfaces patterns like escalations, tool failures, and trust breaks, automated issue detection with alerts via Slack, email, or webhook, automated evaluations that convert real production failures into regression tests, golden-dataset management from validated traces, and MCP-server integration so teams can manage projects and traces from their coding agents. It connects to OpenTelemetry, the Vercel AI SDK, and Claude Code, and can be set up in under five minutes. Latitude offers a free MIT open-source self-hosted edition, a free cloud Starter tier (20K credits/month, 30-day retention, unlimited seats), a Pro plan at $99/month (100K credits, 90-day retention, SOC2 and ISO27001 reports), and a custom Enterprise plan with on-premises or custom-cloud deployment, RBAC, and SAML SSO.
At a Glance
- Category
- Developer Tools
- Pricing
- Free, Freemium, Subscription, Contact for pricing
- Target Market
- Enterprise Developers, AI Engineers, Data Scientists, CTOs
Key Features
- ✓Semantic trace search
Search across 100% of agent traces without sampling to locate failures fast.
- ✓Failure-mode clustering
Groups similar failing traces into actionable issues with states and evaluations.
- ✓Conversation intelligence
Analyzes completed sessions to surface escalations, tool failures, and trust breaks.
- ✓Automated evaluations
Converts real production failures into tests to prevent regressions.
- ✓Automated issue detection
Alerts teams via Slack, email, or webhook when new issues emerge in production.
- ✓MCP and OpenTelemetry integration
Connects to coding agents via MCP and to telemetry pipelines via OpenTelemetry.
Capabilities
Use Cases
- •Production agent debugging
Pinpoint and fix the failure modes behind broken agent behavior before users hit them.
- •Regression prevention
Turn real failures into automated evaluations and golden datasets that guard future releases.
- •Agent quality monitoring
Continuously monitor multi-turn agent sessions and get alerted on new issues.
Ideal For
Best For
- ✓Debugging why AI agents fail in production
- ✓Building golden datasets and automated evaluations
- ✓Monitoring multi-turn agent trajectories
Integrations
Deployment
Market Analysis
Pros
- ✓Open-source with no lock-in and fast setup
- ✓Issue-first workflow rather than raw log dumps
- ✓Full-trace semantic search without sampling
Cons
- ✗Smaller and younger than incumbent observability vendors
- ✗Funding and company details not publicly detailed
Pricing
Open Source
$0
- ✓MIT license
- ✓Self-hosted
- ✓5-minute setup
- ✓Observability, session search, issue discovery
Starter
$0
- ✓20K credits/month
- ✓30-day data retention
- ✓Unlimited seats
Pro
From $99/mo
- ✓100K credits/month
- ✓90-day retention
- ✓SOC2 & ISO27001 reports
- ✓Priority support
Enterprise
Contact for pricing
- ✓Custom credits and retention
- ✓On-prem or custom cloud
- ✓RBAC and SAML SSO
- ✓Uptime/support SLA
Free MIT open-source self-hosted edition plus a free cloud Starter tier; Pro at $99/month; Enterprise custom with on-prem/self-hosted options and no lock-in contracts.
Sources
This page was written from 3 sources, 1 on domains other than latitude.so.
Stay Ahead of the Curve
Weekly enterprise AI insights for technology leaders. No spam, no vendor pitches—unsubscribe anytime.
SubscribeRelated Products
Respan
Observability, evals and an LLM gateway for AI agents in one control plane
Meta Muse Code
Meta's terminal coding agent for large repositories, with persistent background agents and the most aggressive token pricing in the category
Niteshift
The full-stack cloud for coding agents — real environments, verified pull requests
Opik
Open-source tracing, evaluation and guardrails for LLM applications and AI agents