Dust
Platform where companies build and share AI agents connected to their data and tools, with 20+ models, 70+ connectors and enterprise controls
Overview
Dust is a platform where teams build, share and run AI agents that work with company knowledge and tools. Agents connect to more than 70 sources such as Slack, Notion, Google Drive, GitHub, Salesforce and Zendesk, can call MCP servers, and run on your choice of more than 20 models from OpenAI, Anthropic, Google, Mistral and others. Dust calls itself "multiplayer AI": people and agents share work, with human review built in, reusable skills, and permissions that control which data, tools and agents each person can use. Admins get audit logs, usage and cost analytics, SSO and SCIM, US or EU data residency and a single-tenant option, and the company says it holds SOC 2 Type II. Dust was founded by Gabriel Hubert and Stanislas Polu and raised a $40M Series B in May 2026, co-led by Abstract and Sequoia, bringing total funding past $60M.
The Verdict
Who Should Use Dust?
Best For
- Companies that want many teams to build their own agents on internal data
- Ops, support, sales and engineering teams that live in Slack, Notion and Google Drive
- Organizations that want to switch models without switching vendors
- EU companies that need EU data residency
Not Ideal For
- Developers who want full code-level control (a framework such as LangChain fits better)
- Companies that cannot give the platform scoped access to source systems
- Teams looking mainly for enterprise search rather than custom agents
What's Great
- G2 reviewers praise fast agent creation and ease of use
- Many models under one subscription
- Wide connector library for Notion, Drive and Slack threads
- Strong permission model for data, tools and agents
- Customers include Datadog, Clay, Cursor, Vanta and 1Password
Watch Out For
- Answers are only as safe as connector permissions; a review warns of leaks if permissions fail at retrieval
- Setup and governance take work; many overlapping agents can confuse users
- Learning curve for non-technical users on advanced setup
- Some G2 reviewers mention price and limited onboarding help
Pricing
Prices exclude VAT. Business seats cover teams of up to 100 people; credits do not roll over.
View all features & details
Key Features
- No-code agent builder with reusable skills
- 70+ connectors and MCP server support
- 20+ frontier and open models, chosen per agent
- Chrome extension and developer API
- Human review and shared work between people and agents
Security and Admin
- Permissions for data, tools and agents
- Audit logs, usage and cost analytics
- SSO, SCIM, auto-provisioning
- US or EU data residency, single-tenant option, custom retention
Company
- Founded by Gabriel Hubert (CEO) and Stanislas Polu
- Series A of $16M in June 2024, led by Sequoia
- Series B of $40M in May 2026, with Snowflake Ventures and Datadog
How It Compares
| Feature | Dust | Glean | Copilot Studio |
|---|---|---|---|
| Focus | Teams build their own agents | Enterprise search and assistant | Agents inside Microsoft 365 |
| Model choice | 20+ models from several labs | See vendor | See vendor |
| Public self-serve pricing | Yes, from €24/seat | See vendor | See vendor |
| Data residency | US or EU | See vendor | See vendor |
Sources: dust.tt · Dust pricing · TechCrunch (Jun 2024) · 1stPage review
Is this your tool? Add a badge