Dust has emerged as the leading “Operating System” for the AI-native team. While individuals have adopted ChatGPT for personal tasks, organizations have struggled to deploy safe, knowledgeable AI that understands their internal context. Dust solves this by providing a secure platform where companies can build custom AI Agents that have read-access to their proprietary data silos—Notion, Slack, Google Drive, and GitHub—without risking data leakage. By 2026, Dust has become the standard “middleware” for European and US enterprises, allowing non-technical teams to spin up specialized “virtual colleagues” (e.g., a “Customer Support Helper” or “Legal Compliance Bot”) that work alongside them in a shared workspace.
The platform’s defining philosophy is “Model Agnosticism”. Unlike tools locked into a single provider, Dust allows users to swap the underlying “brain” of an agent instantly—using Anthropic’s Claude 3.5 for coding tasks, OpenAI’s GPT-5 for creative writing, or Mistral’s models for European data sovereignty. This flexibility, combined with its robust “Triggers” system (introduced late 2025), has transformed Dust from a passive chat tool into an active automation platform where agents can autonomously wake up, run reports, and ping humans on Slack when they find anomalies.
Read more: 34 US startups to watch in 2026
Core Technology: RAG & The Agent Builder
- The Dust Platform: A no-code interface where teams can create custom agents by simply selecting a model, writing a “System Prompt” (instructions), and checking off which data sources (e.g., “Engineering Wiki” in Notion) the agent can access.
- Live Data Connectors: Real-time, permission-aware synchronizers for Notion, Slack, Microsoft Teams, Google Drive, GitHub, and Zendesk. The AI only “knows” what the specific user asking the question is allowed to see.
- Model Switching: The ability to hot-swap models (GPT-4, Claude 3.5 Sonnet, Gemini Pro, Mistral Large) for specific agents based on cost, speed, or reasoning capability.
- Dust Apps (Developer Platform): A lower-level environment for engineers to build complex, multi-step chains (e.g., “Search database -> Summarize -> Draft Email -> Ask for Approval”) that go beyond simple Q&A.
Business & Market Status
- Funding: Raised approximately $21.5 Million, including a $16 Million Series A in mid-2024 led by Sequoia Capital.
- Traction: By 2026, the platform powers over 80,000 active custom agents across hundreds of high-growth tech companies (like Alan, Qonto, and Payfit), processing millions of internal queries monthly.
- European Leader: Headquartered in Paris, Dust has become a champion for European tech sovereignty, offering robust GDPR compliance and local hosting options that appeal to EU enterprises.
Company Profile
- Founders: Gabriel Hubert (CEO) and Stanislas Polu (CTO). Both are former engineers at Stripe and OpenAI, giving them a unique pedigree in both fintech-grade infrastructure and frontier model research.
- Headquarters: Paris, France (with a presence in San Francisco).
- Key Investors: Sequoia Capital, XYZ Venture Capital, Connect Ventures, Motier Ventures, Seedcamp.
Key Use Cases
| Use Case | Description |
|---|---|
| Support Triage | Support teams build an agent connected to all past Zendesk tickets and Notion help docs to instantly draft answers for new customer queries. |
| Sales Enablement | Sales reps use a “Competitor Tracker” agent that scans public web news and internal Slack channels to provide real-time battle cards against rivals. |
| HR & Onboarding | New hires interact with a “People Ops” agent that can answer “How do I request time off?” or “What is our WFH policy?” by citing the official employee handbook. |
Why It Matters
Dust addresses the “Empty Box” problem of enterprise AI. A generic chatbot is useless if it doesn’t know your business. By solving the hard engineering problems of data synchronization and access control, Dust allows companies to build their own internal “Google” that talks back. It represents the shift from “AI as a tool” (like a calculator) to “AI as a teammate” that shares your context.
