Skip to content

Dust

A workspace for building small AI assistants trained on your team's own documents, connected to Slack and Notion so they answer inside tools you already use.

Assistants & chat Agents

Checked

Open dust.tt

What you'd use it for

Building an assistant that answers questions from your own Slack, Notion and Drive content, without hiring an engineer to wire it up.

  • Chat with the default @dust agent, which answers questions by searching only the data sources you've connected
  • Build a custom agent scoped to one job, such as drafting support replies or summarising meeting notes, without writing code
  • Connect Slack, Notion, Google Drive and dozens of other workplace tools as data sources an agent can search
  • Run several agents in one conversation, switching between your own custom agents and generic model agents like @gpt or @claude
  • Call an agent programmatically from your own code rather than through the chat interface

The verdict

Dust earns its place when a general chatbot doesn't know your company and hiring an engineer isn't worth it. Point it at Slack, Notion and Drive and a non-technical team member can stand up an assistant that answers from your own documents rather than the open web. The credit system means cost tracks actual use rather than a flat seat fee, which rewards a team running it daily and penalises one running it occasionally against large documents. It suits an operations or support team already living inside Slack and Notion more than a solo founder with a handful of files, where the labour of connecting data sources outweighs what it saves.

What it costs

Tier Price For Includes
Free €0 Occasional users, or people trying Dust before committing. Limited usage for exploring the product.
Pro €24/month billed annually, €30/month billed monthly Most team members using an assistant as a daily tool. 8,000 credits per seat per month.
Max €120/month billed annually, €150/month billed monthly Power users running complex, credit-hungry automations. 40,000 credits per seat per month.
Enterprise Custom pricing Organisations with programmatic, high-volume usage. Programmatic usage billed at €0.01 per credit, plus custom contract terms.

Read from Dust's own pricing page on 25 Sep 2026. Vendors change prices without notice.

What you need to start

  • Admin or owner permissions on the Slack, Notion, Drive or other workspace you want an agent to search, to authorise Dust's connector
  • Nothing beyond a Dust account to start chatting with the default agent; the free tier needs no credit card

A few minutes to sign up and start chatting with the default agent; connecting your first data source and getting a grounded answer usually happens inside the same sitting, given admin access to authorise it.

Your first run Connect one data source such as Slack, Notion or Drive as a workspace admin, then ask the default @dust agent a question about it in chat — the first answer pulled from your own documents typically arrives within that same sitting.

Where it breaks

Dust prices in euros by default, and a US-dollar figure could not be forced from this machine even with a US region parameter and English-language headers. Convert at your own bank's rate — the figure above is what the vendor actually charges, not a converted estimate.

Browser-based; connects to Slack, Notion, Google Drive and other workplace tools as data sources for the assistants you build.

First entry in this reference. There is no previous check to compare it against.

When not to use it

Not for a solo operator with only a handful of documents, where the labour of connecting Slack, Notion or Drive and configuring an agent costs more than it saves. Also not the right fit if what you actually want is a handful of fixed, pre-written answers surfaced instantly rather than an assistant that reasons over documents each time — that job belongs to a tool built for it, not this one.

Dust is aimed at a specific gap: a general chatbot like ChatGPT knows nothing about your company, and building a custom one from scratch means hiring an engineer. Dust sits in between — you point it at your Notion pages, your Slack history, your Drive folders, and it builds an assistant that can actually answer “what’s our refund policy” or “how did we price the Q3 deal” because it has read the source material.

The unit of work here is the “agent”: a configured assistant with a specific job — onboarding new hires, drafting support replies, summarising meeting notes — that a non-technical team member can set up without writing code, though the deeper customisation options do reward someone comfortable with prompts and tool configuration.

What the credits actually buy

Pricing runs on a credit system tied to usage rather than a flat per-seat fee. Pro’s 8,000 credits a seat covers regular daily use; Max’s 40,000 is built for someone running the assistant against large documents or complex multi-step tasks repeatedly. The programmatic rate on Enterprise (€0.01/credit) matters if you’re calling Dust’s assistants from your own code rather than a chat interface.

Where it sits next to the alternatives

Against Guru, the split is build-versus-surface: Guru surfaces pre-written answers, Dust builds an assistant that reasons over your documents and can be given a broader task than answering one question.

Alternatives

Get Dust

Buying the tool is not the hard part.

The ebook is about what to hand over, in what order, and how you find out when it goes wrong. One payment, no subscription.

Join the daily email.

Curated AI news for small businesses and their teams. Free, daily. Sent at 7pm SGT. Unsubscribe anytime.