ProductivityPaid Last reviewed August 2026

Dust

A platform for building custom AI agents wired directly into a company's own tools.

Visit Dust Custom enterprise pricing, no public self-serve tier

What is Dust?

Dust lets companies build their own AI agents that connect to internal tools, documents, and data rather than relying on a generic chatbot with no context. It's aimed squarely at enterprises: the agents are configured by a technical team, then used by everyone else through Slack or a simple chat interface.

Dust's whole premise is that a generic AI assistant is only as useful as the context you give it, and most companies never get around to giving it any. Instead of a consumer chatbot, Dust is built as infrastructure — a way for an internal team to connect company data sources and build agents that actually know the company's own processes, tickets, and documents. That makes it a harder sell to individuals and a much easier one to enterprise IT and platform teams, which is exactly where it has focused through 2025 and 2026.

Key features

  • Build custom agents connected to internal data sources
  • Slack and internal-chat-native deployment
  • Connectors for common enterprise tools and document stores
  • Access controls and audit logs for enterprise IT
  • Agent templates for common internal workflows
  • API access for embedding agents into other internal tools

Pros

  • Agents actually reflect a company's real data instead of generic answers
  • Built with enterprise security and access control in mind from the start
  • Flexible enough to cover many different internal use cases

Cons

  • Requires real technical setup work before anyone sees value
  • Pricing is opaque and clearly not aimed at small teams
  • Success depends heavily on how clean the underlying company data is

Best for

Enterprise IT and platform teams building internal AI toolsCompanies with messy but valuable internal documentationOrganisations wary of sending data to a generic public chatbotTeams who want several narrow agents rather than one do-everything bot

Read more