Lunary

About Lunary

Lunary is an observability and prompt-management platform for applications built on LLMs, covering the gap between a prototype that works in a notebook and a deployment you can be accountable for. It logs production and staging traffic as events, gives you searchable traces of what the model was asked and what it returned, tracks cost and latency, and surfaces how real users are actually interacting with a chatbot — which is routinely different from how the team assumed they would. Prompt management is versioned and separated from the application code, so a prompt change does not require a redeploy, and an AI playground lets you test variants against saved data before promoting one. Human review and topic clustering turn the raw log into something a product owner can act on, and Smart Views, custom dashboards and CSV/JSONL export cover the reporting layer. The platform is available self-hosted for teams whose data cannot leave their infrastructure, with SSO, granular access control, PII masking and data-warehouse connectors on the enterprise tier. It is a fit for small AI teams who need real observability without adopting a heavyweight enterprise APM, and the free tier covers personal projects at 10,000 events per month across three projects.

Key Features

Tags