Dust
Build custom AI agents connected to your company's data — without writing code.
Independent overview by new Mantra · updated July 16, 2026
- Category
- IT / Engineering
- Pricing
- Pro $24/seat/mo billed annually ($30 monthly; 8,000 credits); Max $120/seat/mo annually ($150 monthly; 40,000 credits); Enterprise custom
- Implementation
- 1-2 weeks
- Adoption risk
- Low
- Integrates with
- Notion, Slack, Google Drive, GitHub, Confluence, Intercom, Salesforce
What Dust does
Dust lets teams create AI assistants that pull from the solutions they already use — Notion, Slack, Google Drive, GitHub, Confluence, Intercom, and Salesforce. Instead of switching between systems to find information, staff can ask a single agent and get answers grounded in your actual company data.
Each agent can be built around a specific job: answering IT questions, summarising customer feedback, or surfacing relevant documentation during a support conversation. You choose which underlying AI model powers it — Claude, GPT, or Mistral — depending on what fits your needs and preferences.
Implementation is estimated at one to two weeks, which makes it one of the lighter lifts in this category. The platform sits in the IT and engineering function but the use cases — productivity enhancement and data analysis — apply across most business types and industries.
Key capabilities
Custom agent building
Teams can create purpose-built AI agents tailored to specific workflows or departments, without vendor lock-in to a single AI model.
Connected to existing data sources
Agents draw on content from Notion, Slack, Google Drive, GitHub, Confluence, Intercom, and Salesforce, so answers reflect what your organisation actually knows.
Choice of AI model
You can run agents on Claude, GPT, or Mistral, giving you flexibility to match model characteristics to the task or your existing agreements.
Knowledge work automation
Dust is designed to handle repeatable knowledge tasks — retrieving information, drafting responses, analysing data — reducing manual lookup time for teams.
Scalable credit-based pricing
Usage is managed through a credit system, with Pro seats at 8,000 credits and Max seats at 40,000, letting you right-size spend across different user types.
Best for
- IT or engineering teams that want to build and manage internal AI assistants tied to real company documentation
- Businesses already using Notion, Slack, Google Drive, or GitHub as primary knowledge stores
- Teams doing repetitive information retrieval or internal Q&A that could be handled by a trained agent
- Organisations that want model flexibility rather than being locked into one AI provider
- Companies looking for a relatively fast deployment — estimated at one to two weeks — without a large implementation project
Worth knowing
- Pricing is per seat and billed in credits; teams with highly variable usage should map their expected activity against credit limits before committing to a tier
- The Pro plan at $24/seat/month (annual) suits smaller teams, but costs scale with headcount, so larger organisations should model total spend carefully or explore Enterprise pricing
- Dust is strongest when your data already lives in its supported integrations — if your key systems aren't on that list, you may not get full value without additional work
- Building and maintaining agents requires someone comfortable configuring the platform; it is low-code, not no-involvement
- If your organisation has strict data residency or security requirements, these should be validated directly with Dust before proceeding, as vendor site details were not available for this review
Related solutions
Ready to see Dust in action?
Not sure where to start? Our free assessment matches your workflow with the AI solution that fits.