Dust is a team AI-agent platform. Users create custom assistants that can search company data sources, use advanced models, execute actions, and integrate into tools such as Slack, Chrome, Zendesk, APIs, Google Sheets, and Zapier.
It sits between no-code automation and enterprise search. Compared with Glean, Dust is more builder-oriented and lighter-weight. Compared with n8n or Zapier, Dust is more AI-agent-native.
System Verdict
Pick Dust if your team wants useful internal AI agents without building a custom RAG platform. It is strongest for knowledge search plus actions across real workplace systems.
Skip it if you only need generic chat. ChatGPT, Claude, or Gemini will be cheaper and simpler.
Dust’s value is operational context: connected data, team spaces, native integrations, and action-capable agents.
Key Facts
| Core product | Team AI agents and assistants |
| Models | Advanced models including GPT-5, Claude, Gemini, and Mistral |
| Data sources | GitHub, Google Drive, Notion, Slack, and more |
| Integrations | Slack, Zendesk, Chrome Extension, API, GSheet, Zapier |
| Security | SOC 2, zero data retention positioning, private spaces |
| Pricing (verified 2026-06-12) | Pro 29 EUR/user/month excl. tax; Enterprise custom from 100 members |
| Pro limits | Fair-use unlimited messages, programmatic credits, up to 1GB/user data sources, one private space |
| Enterprise extras | SSO (Okta, Entra ID, Jumpcloud), SCIM, US/EU hosting, Salesforce Tool, priority support |
| Best fit | Internal agents for teams |
When to pick Dust
- You need company-grounded assistants. Agents can search selected data sources before answering.
- You need actions, not just answers. Dust agents can execute approved tools.
- You work in Slack. Native Slack integration makes agents reachable where teams already ask questions.
- You want model choice. Dust exposes multiple frontier providers rather than betting on one model.
- You want a builder surface. Non-infra teams can create repeatable agents without owning vector database plumbing.
- You need programmatic usage. API, Google Sheets, and Zapier credits make Dust useful beyond chat-style assistant surfaces.
When to pick something else
- Enterprise-wide search: Glean has deeper enterprise graph and connector positioning.
- General automation: n8n, Zapier, Activepieces.
- One-off chatbot: ChatGPT or Claude.
- Custom app platform: LangGraph or Mastra.
Pricing
Dust publishes a Pro plan at 29 EUR per user per month, excluding tax, with a 14-day trial. As verified on 2026-06-12, Pro starts from one user and includes advanced models (GPT-5, Claude, Gemini, Mistral), custom action-capable agents, core connections, native integrations, privacy/security features, fair-use unlimited messages, programmatic credits, fixed pricing on additional programmatic usage, up to 1GB per user of data sources, and one private space.
Enterprise is custom and starts from 100 members. It adds advanced security and controls, larger storage and file limits, custom programmatic pricing, SSO (Okta, Entra ID, Jumpcloud), flexible payment options, priority support and account management, priority access to new features, US/EU data hosting, SCIM provisioning, and the Salesforce Tool.
The pricing is reasonable for team workflows if agents replace repeated internal support, research, or support-answering labor. It is expensive if used like a casual chatbot.
Buyer fit
Dust is strongest when a team can name repeatable internal jobs: answer support questions from product docs, summarize account history, draft customer responses, search engineering knowledge, pull context from Slack and GitHub, or run an approved action from a shared assistant.
It is weaker when the organization only wants a general chatbot. The value depends on connective tissue: clean source permissions, useful data sources, well-scoped agents, and teams willing to move repeated questions into shared workflows instead of one-off chats.
Rollout checklist
- Start with one department and one high-frequency workflow.
- Connect only the sources needed for that workflow.
- Define which tools an agent may execute and which actions require human review.
- Decide who owns assistant prompts, source hygiene, and regression testing.
- Check whether Pro storage, private space, and programmatic usage limits match the pilot.
- Move to Enterprise review before broad rollout, SSO, SCIM, hosting requirements, or regulated data.
Failure Modes
- Agent quality depends on source data. Bad docs produce bad answers.
- Storage limits matter. Pro includes limited data-source storage per user.
- Action permissions need care. Tool-executing agents require governance.
- Not a full workflow engine. Complex deterministic automation may still belong in n8n or Zapier.
- Enterprise features are custom. SSO, SCIM, hosting, and bigger limits require sales.
- Fair-use language needs review. “Unlimited” messages still depend on fair-use limits and programmatic usage terms.
Recent changes
- Model lineup rechecked (verified 2026-06-12). The Dust pricing page explicitly names GPT-5, Claude, Gemini, and Mistral as available advanced models for Pro and Enterprise.
- Enterprise SSO providers named. Okta, Entra ID, and Jumpcloud are now listed by name on the Enterprise tier, alongside SCIM provisioning, US/EU data hosting, and the Salesforce Tool.
- Pro pricing unchanged. 29 EUR/user/month excluding tax, from 1 user, 14-day trial. The fair-use message policy and 1GB-per-user data-source cap also persist.
Methodology
Last verified 2026-06-12 against Dust pricing, product, and documentation pages. Scoring emphasizes team utility, action-capable agents, integrations, pricing, and enterprise readiness.
FAQ
Can Dust agents search company data? Yes. Dust agents can search selected data sources and use relevant documents to answer.
Does Dust support Slack? Yes. Slack is a native integration.
Is Dust free? Dust offers a trial, but its main self-serve plan is paid per user.
Related
- Category: AI Automation · AI Search
- See also: Glean · n8n · Activepieces · Zapier · LangGraph
