Omni

Omni is an open-source AI agent platform that connects to 15+ work tools including Google Workspace, Slack, Jira, Confluence, and more. Teams can search across their whole organization and get AI-powered answers grounded in real data with source citations.

• Instant answers across all connected apps with cited sources
• Code execution in sandboxed Python and bash environments for data analysis and automation
• Permission inheritance from source systems, so users only see data they already have access to
• Bring your own LLM with support for Anthropic, OpenAI, Gemini, Azure, or self-hosted models via vLLM
• Faceted search with filters by source, date, type, and author
• MCP support for connecting Model Context Protocol tools and clients

Omni is fully self-hosted on your infrastructure with Apache 2.0 licensing. It uses PostgreSQL with ParadeDB for full-text and semantic search, so you don't need Elasticsearch. Deploy via Docker Compose for development or Terraform modules for production on AWS or GCP. The Connector SDK lets you build custom integrations in Python or TypeScript for any internal or external data source.

Each connector runs in its own lightweight container. The code sandbox has no access to internal services or the internet.