Onyx

About Onyx

Onyx is an open source AI teammate that connects any large language model to a team's private data. It offers a chat-based UI for asking questions, running research across web and internal documents, and creating agents that perform actions on behalf of users.

Review

Onyx aims to bridge the gap between generic AI assistants and the specific context teams need by integrating with internal docs, Slack, and cloud storage. The platform combines search, agent actions (MCP), and the ability to use custom LLMs, which makes it useful for practical, work-focused tasks rather than casual chat.

Key Features

• Connect any LLM to your team's data sources for contextual responses.
• Build custom agents with actions and MCP to automate workflows.
• Search both internal documentation and the web for deep research capabilities.
• Support for private, open source deployment alongside hosted options.
• Data connectors and ad-hoc analysis (example: read CSVs, compute metrics like MRR growth).

Pricing and Value

The project highlights free options and a launch promotion that includes a multi-month free period for some users. Because the core is open source, teams can self-host at no licensing cost, while hosted plans are available for those who prefer managed infrastructure. The value proposition centers on reducing friction between models and real team knowledge, which can save time on research and meeting prep compared with manual document searches.

Pros

• Open source with options for private/self-hosting, which suits teams with strict data requirements.
• Directly links LLMs to internal context, improving relevance for work tasks.
• Agent actions and MCP enable automation of routine processes and note preparation.
• Combines web search with internal search for broader research capability.

Cons

• Early-stage product may lack polish and some integrations compared with mature commercial offerings.
• Self-hosting requires technical setup and maintenance, which can be a barrier for smaller teams.
• Certain advanced features may be gated behind hosted plans or require configuration effort.

Onyx is best suited for product, research, and operations teams that need AI tied to internal knowledge and prefer open source or private deployments. Organizations that want an extensible chat interface with agent capabilities and the option to run their own LLMs will find it particularly useful.