Skip to content
SPECIFICTION
Get listed

Latitude

latitude.so

From $99/mo Free plan Open source WebLinuxAPISelf-hosted
Visit latitude.so
Spec score 9.2 #2 of 28 in AI Prompt Generators

What it is

Latitude is an open-source, MIT-licensed platform for monitoring and improving production AI agents. It records agent interactions as spans and groups related traces into sessions for reviewing multi-turn conversations. Observe tools expose model providers, models, tokens, latency, cost, errors, tool calls, user activity, and agent memory. Semantic search, behavior clustering, and signals help locate recurring failures; monitors can watch signals, saved searches, or raw traffic and notify users in-app, by email, or through Slack. Failing traces can become datasets for local simulation and regression testing, while evaluations monitor live traffic for regressions. Latitude offers managed cloud and self-hosted deployment, including Docker Compose and Kubernetes options. Its setup guide names TypeScript and Python SDKs and recommends an OpenTelemetry exporter for other languages. The free Starter plan includes 20K credits per month and 30-day data retention. Pro costs $99.00 per month and includes 100K credits monthly and 90-day retention. Latitude says hosted customer data and managed inference are processed in AWS eu-central-1 in Frankfurt, Germany.

Spec sheet

What it does
Latitude monitors production AI agents, identifies recurring failures, and can dispatch coding agents to implement fixes.Read 15 Oct 2026 · latitude.so
Purpose
Latitude is an open-source, MIT-licensed platform for improving production AI agents by capturing agent traffic, surfacing repeated failures, and helping teams monitor and fix them.Read 3 Oct 2026 · docs.latitude.so
Deployment
Latitude is available as managed cloud or self-hosted software, with documented single-machine Docker Compose and Kubernetes cluster deployments.Read 3 Oct 2026 · docs.latitude.so
Integrations
Documented integrations include OpenAI, Anthropic, Amazon Bedrock, Vercel AI SDK, LangChain, Claude Code, Slack, GitHub, Linear, webhooks, and OpenTelemetry.Read 15 Oct 2026 · github.com
Security
Latitude states that it has SOC 2 Type II certification, TLS 1.2-plus in transit, AES-256 at rest, SAML SSO, GDPR compliance, European data residency, and exportable audit logs.Read 15 Oct 2026 · latitude.so
Compliance
Latitude's security documentation states it is SOC II certified and says ISO 27001 audit work and GDPR readiness review were in progress, with completion expected by the end of May 2026.Read 3 Oct 2026 · docs.latitude.so
License
Latitude is fully open source under the permissive MIT license.Read 15 Oct 2026 · github.com
Agent dispatch
When a signal appears or escalates, Latitude can send context and sample traces to Claude Code or Cursor and open a pull request.Read 15 Oct 2026 · latitude.so
API and extensibility
Latitude exposes an API, CLI, MCP server, SDKs, webhooks, and coding-agent skills for programmatic control.Read 15 Oct 2026 · docs.latitude.so
Company profile
LinkedIn lists Latitude as a privately held software-development company founded in 2022, headquartered in San Francisco, with 2–10 employees and an additional Barcelona location.Read 15 Oct 2026 · linkedin.com
Data minimization
Latitude says its SDKs redact common security-sensitive attributes by default and customers can configure additional redaction patterns.Read 3 Oct 2026 · docs.latitude.so
Datasets and regression
Failing traces can become datasets for local simulation and regression testing before deployment.Read 15 Oct 2026 · docs.latitude.so
Documentation depth
The documentation covers tracing, providers, frameworks, agent harnesses, observability, signals, experiments, monitors, datasets, security, deployment, API, MCP, CLI, integrations, and development.Read 15 Oct 2026 · docs.latitude.so
Experiments
Experiments compare traffic slices to measure differences between releases, models, or user cohorts.Read 15 Oct 2026 · docs.latitude.so
Failure clustering
Latitude groups similar failing traces into signals with examples, trends, affected users, and lifecycle states.Read 15 Oct 2026 · docs.latitude.so
Failure discovery
Semantic search, behavior clustering, and signals help teams find and track recurring production failure patterns.Read 3 Oct 2026 · docs.latitude.so
Human annotations
Users can annotate traces, spans, and outputs, with annotations feeding structured signals and evaluations.Read 15 Oct 2026 · latitude.so
Legal entity
The terms identify the provider as LATITUDE DATA, S.L., incorporated in Spain with Tax ID B09930660 and a Barcelona registered address.Read 15 Oct 2026 · latitude.so
Limit: ecosystem maturity
MakerStack describes Latitude’s community as younger and smaller than Langfuse’s, with fewer integrations and tutorials.Read 15 Oct 2026 · makerstack.co
Limit: traffic and telemetry
DiscoverAI cautions that useful signals need representative production traffic, sensitive telemetry requires redaction and retention controls, and automated clustering needs human calibration.Read 15 Oct 2026 · discoverai.tools
Limit: usage costs
MakerStack says credit-based Pro metering and the $20 charge per additional 10K credits can become expensive for high-traffic agents.Read 15 Oct 2026 · makerstack.co
MCP
Its MCP server lets coding agents inspect and manage Latitude workspaces, projects, traces, annotations, and datasets.Read 15 Oct 2026 · github.com
Monitoring
Monitors can watch signals, saved searches, or raw traffic and notify users in-app, by email, or in Slack.Read 3 Oct 2026 · docs.latitude.so
Monitors and alerts
Monitors watch signals, searches, or raw traffic and notify in the app, by email, or through Slack.Read 15 Oct 2026 · docs.latitude.so
Notable release
The changelog records Agent Score launching on September 23, 2026 as a 0–100 score based on five dimensions and evidence thresholds.Read 15 Oct 2026 · latitude.so
Observability
It records traces and spans for LLM calls, tool calls, retrieval, tokens, cost, latency, and errors, grouping related traces into sessions.Read 15 Oct 2026 · docs.latitude.so
Praise: agent-focused tracing
Independent reviewers repeatedly praise multi-turn session traces that include tool calls and complete execution paths.Read 15 Oct 2026 · makerstack.co
Praise: issue-to-evaluation workflow
Independent reviews praise the workflow that turns production failures into tracked issues and reusable automated evaluations.Read 15 Oct 2026 · makerstack.co
Privacy retention
The privacy policy says retention is 30 days for Starter, 90 days for Pro, and customized for Enterprise.Read 15 Oct 2026 · latitude.so
Rating on DiscoverAI
8.0 out of 10 from 1 editorial reviewRead 15 Oct 2026 · discoverai.tools
Rating on MakerStack
7.7 out of 10 from 1 editorial reviewRead 15 Oct 2026 · makerstack.co
Regression workflow
Datasets turn failing traces into cases for local simulation and regression testing, while evaluations can monitor live traffic for regressions.Read 3 Oct 2026 · docs.latitude.so
SDKs
The project provides TypeScript and Python telemetry SDKs and supports any OpenTelemetry-compatible runtime.Read 15 Oct 2026 · github.com
Self-hosting
Self-hosting is supported from a Docker Compose single host to a Kubernetes cluster using Helm, with Railway deployment also documented.Read 15 Oct 2026 · github.com
Semantic search
Search supports semantic meaning, exact text, and metadata filters across production traces.Read 15 Oct 2026 · docs.latitude.so
Support and audience
The pricing page describes Starter for teams establishing AI infrastructure, Pro for collaborative AI product teams scaling up, and Enterprise for high-volume products or infrastructure deployment; support ranges from general Slack channel support to dedicated support.Read 3 Oct 2026 · latitude.so
System requirements
The development setup requires Git, Docker and Docker Compose, Node.js, and pnpm; tmux and tmuxinator are optional recommendations.Read 15 Oct 2026 · docs.latitude.so
Telemetry SDKs
The setup guide names TypeScript and Python SDKs and recommends an OpenTelemetry exporter for other languages.Read 3 Oct 2026 · docs.latitude.so
Tracing
Latitude traces agent interactions as spans and groups related traces into sessions for reviewing multi-turn conversations.Read 3 Oct 2026 · docs.latitude.so

Plans and prices

Starter Free 20K credits/month · 30d data retention · unlimited seats · general-channel Slack support Read 15 Oct 2026 · latitude.so
Pro $99 a month /month 100K credits/month · 90d data retention · unlimited seats · priority support · extra credits $20 per 10K · SOC2 & ISO27001 reports Read 15 Oct 2026 · latitude.so
Enterprise Priced on request custom credit volume · custom retention · custom on-premises or cloud deployment · RBAC · team training · SAML SSO · SOC2 & ISO27001 reports · uptime/support SLA · dedicated support Read 15 Oct 2026 · latitude.so

Every Latitude plan and price →

Strong points and limits

Strong points

  • Groups agent traces into reviewable sessions
  • Monitors can notify in-app, by email, or Slack
  • Failure traces can become regression datasets
  • Offers managed cloud and self-hosted deployment
  • Free Starter includes 20K monthly credits

Limits

  • Starter data retention is 30 days
  • Pro costs $99.00 per month
  • Extra Pro credits cost $20 per 10K

Who it suits

Latitude suits teams that need to observe production AI-agent traffic and turn recurring failures into regression cases. Its managed cloud and self-hosted options serve teams with different deployment preferences.

Verdict

Latitude links agent observability with workflows for identifying and checking recurring failures. Compare the Starter and Pro credit and retention limits, or review the custom Enterprise terms if those matter to your deployment.

Where Specifiction places it

Alternatives to LatitudeAll alternatives →