Lunary

Lunary is a self-hosted LLM observability and prompt management platform for teams building chatbots and AI agents. It brings production traces, user conversations and prompt versions into one place so developers can investigate errors and teams can assess response quality. You can host it in your own infrastructure or use its cloud service.

Agent traces record prompts, results and error stack traces, with search and filters for finding specific interactions. Chat replays show how people use an application, while analytics track model usage, costs, latency and satisfaction. Topic classification reveals recurring questions, and custom dashboards let teams focus on the measures relevant to their product. Alerts flag agents that aren't performing as expected.

Prompt templates give developers and non-technical teammates a shared place to edit prompts, track versions and run A/B tests. A playground lets teams try prompts with different models without writing code. Human reviews, response scoring and feedback tracking support quality checks, and teams can label recorded data for fine-tuning. Lunary also handles text, audio and image interactions.

Enterprise self-hosted deployments run through Docker or Kubernetes, including within a VPC, so teams can keep monitoring data in their own environment. The hosted service runs in Lunary's cloud. Integrations include OpenAI and LangChain. PII masking helps protect personal information in captured data, while role-based access controls and SSO govern who can access it. Self-hosting is governed by Lunary's deployment and licensing terms.