Onyx

Onyx builds the open-source application layer for large language models. Its core product is an AI chat platform that connects to a company's documents, applications and people to produce grounded, cited answers drawn from internal knowledge. The system ingests and syncs from workplace tools such as Google Drive, Slack, Google Docs, Notion and GitHub, then combines hybrid search, retrieval-augmented generation, contextual retrieval and LLM-based knowledge graphs to retrieve and attribute information. It also supports the Model Context Protocol (MCP), code execution, deep research and custom AI agents.

The platform is available both as a self-hosted deployment and as Onyx Cloud, a hosted option. According to the company, it is trusted by more than 1,000 teams, spanning organisations from startups to global enterprises across industries. Onyx's stated priorities include security, transparency and customisation, consistent with its open-source approach.

The engineering work touches several active areas of applied machine learning: retrieval-augmented generation, hybrid search, contextual retrieval, LLM-based knowledge graphs, agent design and data ingestion and syncing connectors, alongside enterprise search and knowledge management.