Dust AI

Dust is an enterprise AI agent platform designed around a simple premise: AI assistants are most useful when they actually know your company's data. Rather than building chatbots that only access public information, Dust connects to your internal data sources — Slack, Notion, Google Drive, GitHub, Confluence, Intercom — and lets you build custom AI agents that can search, summarize, and act on that organizational knowledge.

Data Connections: The Foundation

Dust's real value starts with its data connectors. Connect your Slack workspace, and agents can search message history across channels. Connect Google Drive, and they can read and reference documents. Connect GitHub, and they can pull code context. Connect Notion, and they can navigate your knowledge base. The platform syncs data automatically, so agents always work with current information rather than stale snapshots.

Critically, Dust maintains a zero-data-retention policy with its LLM providers. Your company data passes through models for inference but is never stored or used for training. This is backed by SOC 2 Type II certification and GDPR compliance — not just marketing claims but audited security controls.

Custom Agent Builder

Building agents in Dust doesn't require coding. You configure agents by defining their instructions (system prompt), selecting which data sources they can access, choosing which LLM to use (GPT-4, Claude, Gemini, Mistral), and optionally giving them the ability to execute actions. Agents can be specialized — a sales agent that knows your CRM data, a support agent that searches your help docs, an engineering agent that understands your codebase. The platform handles RAG retrieval, context management, and response generation.

Native Integrations

Dust agents deploy where your team already works. The Slack integration turns agents into conversational bots in any channel. The Chrome Extension provides AI assistance on any webpage. Zendesk integration enables AI-powered customer support. Programmatic access via API, Google Sheets, and Zapier connectors extends agents into automated workflows. This distribution matters — agents that live where work happens get used; agents locked in a separate UI don't.

Model Flexibility

Every agent can use the best model for its task. Complex reasoning tasks get GPT-5 or Claude. Simpler queries use faster, cheaper models. Dust handles the routing and context formatting, so you don't need to manage model-specific prompt templates. This multi-model approach keeps costs reasonable while maintaining quality where it matters.

Pricing and Scale

Dust's Pro plan costs €29/user/month with unlimited messages (fair use limits apply), access to advanced models, custom agents, all data connections, and 1GB/user of data source storage. The 15-day free trial provides full access to evaluate fit. Enterprise plans (100+ users) add multiple workspaces, SSO, advanced security controls, user provisioning, and larger storage limits. Custom pricing is required for Enterprise.

Dust has grown from approximately $7M ARR to targeting $20M+ ARR through 2025, with 1,000+ enterprise customers and expanding toward 5,000+. The company is backed by significant venture funding and led by former Stripe engineers who understand enterprise software requirements.

Competitive Position

Dust competes with Glean (enterprise search), Cassidy (AI agent builder), and Stack AI (no-code AI workflows). Against Glean, Dust offers more agent customization but less sophisticated search ranking. Against Cassidy and Stack AI, Dust provides stronger native integrations (Slack, Zendesk) and simpler setup but less flexibility for complex multi-step workflows. For teams wanting full code control over their AI agents, frameworks like LangChain or CrewAI offer more flexibility at the cost of more engineering effort.

For an overview of enterprise AI agent options, see our guide on best AI agents for solopreneurs and how to choose your first AI agent.