Dust AI

Dust ships managed connectors for Slack, Notion, Google Drive, GitHub, Confluence, Intercom, and Zendesk that sync continuously rather than via one-time imports. The platform handles chunking, embedding, and retrieval automatically, so agents always work with current data without engineering teams maintaining custom pipelines.

Build agents by writing natural-language instructions, selecting which data sources they can read, and choosing a model — no code required. Non-technical team leads can create and iterate on department-specific agents (sales, support, HR) without filing engineering tickets, and agents can be cloned and adapted across teams.

Each agent can use a different model — GPT-4o, Claude, Gemini, or Mistral — based on the task. Dust handles model-specific prompt formatting and context windows, so you get premium reasoning where it matters and cheaper inference for routine queries without managing per-provider integrations yourself.

Agents deploy directly into Slack, a Chrome Extension, Zendesk, Google Sheets, and Zapier, plus a programmatic API. This is decisive in practice — agents that live inside Slack and Chrome get used daily, while agents stuck behind a separate UI typically see usage decay within weeks of launch.

SOC 2 Type II certified and GDPR compliant, with a zero-data-retention policy enforced contractually with LLM providers. Enterprise plans add SSO, user provisioning, multiple workspaces, and advanced access controls — the controls security review teams actually ask about, not just badges on a marketing page.