Dust is a multiplayer AI platform where people and agents work from the same company context. Teams connect more than 70 sources, including Slack, Notion, Google Drive, GitHub, Salesforce and Zendesk, then build agents on 20-plus frontier and open-source models from providers such as OpenAI, Anthropic, Google and Mistral. Agents can answer from the full picture across systems and take actions, with human review where needed. Administrators control permissions for data, tools and agents, and use audit logs and analytics to track adoption, usage, costs and ROI.
Dust is built for operations-minded teams in mid-size and large companies: engineering, customer support, sales, marketing and data groups, plus the AI leads and IT security staff who govern them. Over 3,000 organizations use it, and GTM engineers at companies such as Clay describe building agents that do work their teams once did by hand. Security reviewers get SOC 2 Type II certification, SCIM-synced groups, SSO, US or EU data residency and optional dedicated single-tenant deployment.
Concrete scenarios appear by department. Engineering uses agents for debugging with context from docs and past issues, automated code reviews, incident response runbooks and continuous documentation updates from code changes. Support teams triage and route tickets and draft replies grounded in a knowledge base. Sales and marketing teams pull context from CRM records and shared documents to prepare outreach and content, while data teams run analysis flows. Because skills, tools and conversations are shared, work can move between a person and an agent and back again within one workspace.
