Onyx is an open-source generative AI and enterprise search platform that helps organizations provide their teams with a unified, natural-language interface to large language models while connecting those models to the company's internal knowledge, documents, and applications. Formerly known as Danswer, Onyx serves as a centralized AI workspace where users can chat with LLMs, search across organizational data, build custom AI agents, and automate workflows, all within a secure and permission-aware environment. Onyx falls into the category of enterprise AI platforms, sitting at the intersection of AI-powered chat, knowledge management, and enterprise search. It is designed for a broad audience, from individual developers running it on a laptop to large organizations deploying it on-premise or in the cloud. Common users include engineering teams seeking fast access to internal documentation, sales teams researching customer histories and pipeline data, support teams resolving tickets with accurate knowledge, and operations teams streamlining company-wide information access. The platform connects to over 40 enterprise data sources, including Google Drive, Slack, Confluence, Salesforce, GitHub, Jira, and many others, syncing content in real time and automatically respecting document-level access controls from the source applications. Users interact with this knowledge through a conversational chat interface or a dedicated search experience, and every response is grounded in retrieved documents with inline citations so answers can be verified. Key features and value propositions include: Model-agnostic AI chat. Onyx works with any LLM, including OpenAI GPT, Anthropic Claude, Google Gemini, Meta Llama, DeepSeek, and self-hosted models via Ollama or vLLM, giving organizations full flexibility over their AI provider. Advanced retrieval and accuracy. The platform combines hybrid search (semantic plus keyword), advanced retrieval-augmented generation (RAG), contextual retrieval, and LLM-based knowledge graphs to minimize hallucinations and surface highly relevant results. Custom AI agents and actions. Administrators and users can create tailored AI agents with specific instructions, scoped knowledge sets, and the ability to take actions in external systems via MCP or OpenAPI integrations. Security-first, self-hostable architecture. Onyx can be deployed fully air-gapped with no data sent to third parties. It supports SSO (OIDC, SAML, OAuth2), role-based access control, and encryption of credentials, making it suitable for highly regulated industries. Advanced productivity tools. Built-in capabilities such as deep research, code interpreter, web search, image generation, file uploads, and persistent memory allow teams to handle complex analytical and creative tasks directly within the platform. Onyx is available as a free, MIT-licensed Community Edition for self-hosting, as well as a managed Onyx Cloud offering and an Enterprise Edition with additional features like usage analytics, enhanced permissioning, and dedicated support. It is trusted by organizations such as Netflix, Ramp, Thales Group, Bitwarden, Roku, and Applied Intuition.