Latitude

Latitude is an open-source AI prompt engineering, deployment, and evaluation platform (often described as an observability and reliability platform for AI agents) built for developers, product managers, and cross-functional teams shipping production-grade LLM features.

Key capabilities include:

• Prompt management with version control, collaborative editing, and support for advanced prompt structures (variables, conditionals, etc.).
• Playground for interactive testing and iteration of prompts with different models, inputs, and parameters.
• AI Gateway to deploy prompts as API endpoints that stay current with published changes.
• Datasets, evaluations, and experiments for batch testing and regression checks (LLM-as-judge, programmatic rules, or human review).
• Traces and observability that automatically capture prompt and model interactions from real usage.
• Prompt optimization and reliability loop that turns production failures into automatic improvements.

Pricing: Team: $299/month. Scale: $899/month. Enterprise: Custom (includes self-hosting options). A 30-day free trial is available for cloud plans. Self-hosted version is free.

What makes it notable: It closes the loop between prompt development and production through a defined reliability workflow (design, test, deploy, trace, evaluate, improve), with self-hosting guidance plus SDK and HTTP API options for integrating into existing stacks.