Dust is a platform for building custom AI assistants that are grounded in your company's specific data and connected to your internal tools. You configure data sources (Notion, Confluence, Slack, Google Drive, GitHub), define what the assistant is allowed to see, and deploy it to your team via a web interface, Slack integration, or API. Beyond a single assistant, Dust supports multi-agent workflows: a research agent pulls from documentation, a coding agent has access to your GitHub repository, and an orchestrator routes questions to the right agent based on topic. The agent builder is low-code: you configure agents through a UI with a selection of built-in tools and data connectors rather than writing agent logic from scratch. Dust targets the middle ground between 'bolt ChatGPT onto your data' and 'build a full custom RAG system': it provides real data security and access controls (employees only see data they're authorized to access) while being deployable without an engineering team. Pricing is per seat and targets teams of 5-500 people.

What the community says

Dust is discussed favorably in product and operations communities as a practical tool for giving teams AI that knows their specific company context. Users appreciate the permission system that respects existing data access controls rather than giving all employees access to all company data. Some teams find the customization more limited than building their own RAG system, and note that the quality of answers depends heavily on how well-organized the source data is. The Slack integration is consistently praised for meeting employees where they already work.

See alternatives to Dust →