Quick verdict
Bottom line up front: Dust lets a whole company build its own AI agents on top of the tools it already uses — Slack, Notion, Google Drive, GitHub, Salesforce and 100+ other integrations — with a choice of frontier models per agent. Pricing is per seat and metered in credits: a free seat with 500 lifetime credits, Pro at €24 per user per month billed yearly (€30 monthly) with 8,000 credits, and Max at €120 yearly (€150 monthly) with 40,000. It is one of the strongest “AI for the whole company” platforms, provided your knowledge actually lives in connectable tools.
Research-based review — prices checked October 22, 2026. Dust lists prices in euros, excluding VAT. We do not claim hands-on testing.
What does Dust do?
Dust is a workspace where teams build custom AI agents without code. Connect your company data sources — Slack, Notion, Google Drive, GitHub, Gmail, Zendesk, HubSpot, Jira, BigQuery, Snowflake, Microsoft 365 and Google Workspace among them — and the data stays synced automatically instead of going stale in uploaded files. Anyone can then build an agent for a job: a support agent drafting answers from tickets and help docs, a sales agent researching prospects and updating the CRM, an engineering agent searching the codebase, or a deep-research agent running multi-step investigations across sources.
Agents are model-agnostic: each can run on a different frontier model from OpenAI, Anthropic, Google or Mistral, so a quick lookup does not pay for the heaviest reasoning model. Agents can also act, not just answer — updating records, creating issues and tickets, posting to channels — and run on schedules or triggers, with MCP support for custom tools. People use them where they already work: Slack or Teams threads, a Chrome extension, or Zendesk. Shared Pods keep persistent project context, and Frames turn output into dashboards. Dust cites SOC 2 Type II, HIPAA and GDPR compliance.
How much does Dust cost?
| Seat type | Price | What you get |
|---|---|---|
| Free | €0 | 500 lifetime credits; a seat inside a Business workspace alongside paid seats |
| Pro | €30/user/month monthly, or €24 billed yearly | 8,000 credits per seat per month; the standard seat for regular team members |
| Max | €150/user/month monthly, or €120 billed yearly | 40,000 credits per seat per month, for heavy automations, deep research and tool-heavy work |
| Enterprise | Custom | Workspace-pooled credits, unlimited connectors and MCP servers, SCIM, audit logs, single-tenant deployment options |
Credits are the meter to watch: a lightweight chat costs few credits, while a deep-research run with many tool calls costs many more, so two people on identical Pro seats burn their allowance at very different speeds. Seat types can be mixed in one workspace — most staff on Pro, power users on Max, occasional users on Free.
What are Dust’s strengths?
- Agents grounded in live company data: connectors sync automatically across 100+ tools, so answers cite current docs, tickets and threads — not last quarter’s upload.
- No model lock-in: pick the model per agent and switch as better models ship, instead of betting the company on one vendor.
- Built for non-technical builders: department teams create and iterate on their own agents without an engineering queue.
- Agents that act: scheduled runs, triggers and write-backs to CRMs and trackers move it past chatbot territory into actual workflow automation.
- Enterprise posture: SOC 2 Type II, HIPAA and GDPR, with SSO, SCIM, audit logs and pooled credits at the top tier.
What are Dust’s weaknesses?
- The free seat is a taster, not a tier: 500 lifetime credits run out quickly, and there is no monthly refresh.
- Credit metering blurs the budget: per-seat pricing looks predictable until heavy agents make usage per seat vary wildly.
- Value depends on your stack: a company whose knowledge sits in one tool — or in people’s heads — gets far less from Dust than the demo suggests.
- Internal focus: Dust is built for employees, not customer-facing website or WhatsApp chatbots, and it is cloud-only with no self-hosting.
Who is Dust for?
Dust fits mid-size and larger teams — roughly 30 people and up — whose work is scattered across Slack, Notion, Drive, a CRM and a warehouse, and who want department-specific agents rather than one generic chatbot. Solo users and tiny teams will usually find a single AI subscription simpler and cheaper, and companies shopping for a public support bot should look elsewhere.
Frequently asked questions
Is Dust free?
There is a free seat with 500 lifetime credits inside a Business workspace. It is enough to evaluate Dust, not to run it day to day — paid Pro seats start at €24 per user per month billed yearly.
Which AI models does Dust use?
Dust is model-agnostic: agents can run on frontier models from OpenAI, Anthropic, Google, Mistral and others, and each agent can use a different model.
How is Dust different from ChatGPT Team or Copilot?
Those give everyone the same general assistant inside one ecosystem. Dust lets teams build many specialized agents over live data from 100+ tools, with actions, schedules and shared workspaces.
Verdict
Dust is what “AI adoption” looks like when it works: not one chatbot, but dozens of small agents that know where the company’s knowledge lives and can act on it. The credit meter and the per-seat price demand a pilot with real usage before a company-wide rollout — start one department on Pro, watch the credit burn for a month, then decide who actually needs Max.