Onyx

Onyx is an AI search and chat platform for teams with information scattered across workplace apps. It gives employees and AI agents answers grounded in company documents.

• Company search: Search indexed knowledge from GitHub, Google Drive, Slack and Confluence, with source permissions checked when answering questions.
• Custom agents: Give agents selected knowledge, instructions and actions, including access to external applications through MCP.
• Research and artifacts: Build detailed reports, supplement internal knowledge with web search, and generate downloadable documents and graphics.
• Workplace access: Ask questions through web and desktop apps, Slack, Discord or a Chrome extension; an MCP server supplies company context to Claude Code, Codex and other compatible assistants.

Onyx can run in your own cloud or on bare metal, including a fully air-gapped deployment. It supports Docker and Kubernetes. With local embedding models and LLMs, processing and model traffic can stay within your infrastructure. Supported backends include Ollama, LiteLLM and vLLM; choosing Anthropic, OpenAI or Gemini sends model requests to that provider.

The free Community Edition uses the MIT license and includes core search, chat, agents and actions. Onyx Cloud is a separate hosted offering. Lite mode provides chat and agents with a smaller resource footprint, but it can't index documents.