The open source AI teammate you've dreamed of
AI for real work, not just chat. Connect any LLM to your team's data. Build AI agents with actions & MCP. Search internal docs & the web. Deep research everything. All private and open source.
Onyx connects to 40+ apps (Slack, Google Drive, GitHub, Gmail, etc.) and uses a 6-stage retrieval pipeline with hybrid search, advanced RAG, and LLM-based knowledge graphs to deliver grounded answers with citations. Benchmarks show 64-76% win rates against ChatGPT Enterprise, Claude Enterprise, and Notion AI on real workplace questions.
Onyx offers flexible configuration for LLM providers, search settings, connected apps, and access controls. Supports custom AI agents, MCP/OpenAPI actions, developer APIs, and hook extensions, enabling teams to build tailored solutions without vendor constraints.
Onyx provides RBAC, permission inheritance, OIDC/SAML SSO, and usage dashboards for enterprise-wide rollout. Supports on-premise and region-specific deployments with dedicated support and SLAs, enabling secure access to GenAI for every team member.
Onyx delivers 30x ROI at Ramp by answering thousands of questions weekly and saving ~30 minutes per user per day. Features include deep research, code interpreter, web search, and chat sharing to accelerate information retrieval and collaboration across departments.
Onyx is open-source and can be self-hosted on any infrastructure, maintaining full transparency and control over company data. Includes SOC 2 Type II, GDPR compliance, granular permissions, encryption of secrets, and permission inheritance to ensure data security and access control.