Dust, a startup that helps companies build AI agents that work across teams rather than for individual employees, has raised $40 million in a Series B round. The company was co-founded by Gabriel Hubert and Stanislas Polu, who met at Stanford in 2007 and have worked together ever since. Their previous company, a data analytics firm called TOTEMS, was acquired by Stripe in 2014.
The core problem Dust is addressing is straightforward. Most AI tools in the workplace today are built for one person at a time. An employee opens a chat window, asks a question, gets an answer, and that answer stays with them. Nothing is shared, nothing is stored in a way that benefits the next person. A 2026 survey by Writer found that 79% of organizations struggle to turn AI adoption into real business value. The bottleneck is not the technology itself. It is the lack of a shared environment where people and AI systems can work with the same information toward the same goals.
The Series B was led by Abstract and Sequoia, with Snowflake Ventures and Datadog also participating. Total funding now stands at over $60 million.
Dust uses the term "single-player AI" to describe how most companies currently use these tools. Each employee has their own AI assistant, their own conversation history, their own outputs. A sales rep spends an hour researching a client. The next day, a colleague working on the same account starts over from zero. Work gets repeated. Knowledge disappears. The gains stop at the individual level.
The numbers reflect this. A 2026 Writer survey found that 97% of executives say they are seeing some benefit from AI, yet only 29% report significant organizational ROI. That gap, between what AI delivers for individual users and what it delivers for the business as a whole, is the problem Dust is designed to solve.
"Most AI platforms are stuck in single-player mode: one person, one chatbot, one task," said Ramtin Naimi, General Partner at Abstract. "Dust is multiplayer. AI Operators inside companies like Datadog and 1Password don't just use Dust; they build agents that collaborate across teams, learn from every interaction, and rewire how the entire company works. That's a new operating model and category. That's why we participated in this round."
The new capital will go toward three areas. The first is making agents self-improving, so they get better at their tasks over time based on actual team usage and feedback, without requiring manual updates. The second is deepening the collaboration tools, giving humans and agents the same access to shared projects, files, and context so they can genuinely work side by side. The third is building out enterprise infrastructure, including finer-grained permission controls, cost monitoring, and audit trails, to make the platform reliable for large organizations.
Alongside these product priorities, Dust has begun identifying a new type of internal role it calls the AI Operator. These are not engineers or data scientists. They are people working in operations, sales, marketing, or support who take responsibility for building and running AI systems within their own teams. Dust positions itself as the platform designed specifically for this kind of hands-on, non-technical builder.
"This is a century-defining transformation, and we're only in year three," said Gabriel Hubert, Co-Founder and CEO of Dust. "What will transform the way we work isn't the next best model or assistant. It's going to be a completely new type of system that gives humans and agents shared, governed access to the same information and capabilities so that they become true collaborators, working with the same context, notifications, artifacts, and goals to compound organizational impact. This is what we call multiplayer AI, and this is what we're building at Dust."
The company incorporated in February 2023. Before founding Dust, Hubert had served as Chief Product Officer at Alan, a French health insurance company. Polu joined OpenAI as a research engineer on Greg Brockman's team, where he co-authored papers on AI reasoning with Ilya Sutskever. He left in September 2022 convinced that the models were already capable enough to drive real economic value, but that the product layer needed to deploy them at an organizational level simply did not exist yet. Both founders had spent five years at Stripe following the 2014 acquisition of TOTEMS.
Dust's platform connects to more than 100 data sources and business tools. It does not rely on a single AI model provider, so customers are not tied to one vendor's technology. The system supports document creation, spreadsheet analysis, data visualizations, and other file-based work, and includes memory features that help agents improve based on how teams actually use them. Security and compliance are built in. The platform is SOC 2 Type II certified and GDPR compliant, with data residency options in both the EU and the US.
Adoption figures reported by the company are notable. More than 3,000 organizations use Dust, with monthly active usage above 90% and weekly active usage above 70%. In 2025, the company reported 240% net revenue retention and zero churn. Over 300,000 agents have been deployed across the platform. Among its customers, Clay used Dust to support a fourfold increase in team size without adding proportional headcount in enablement. Persona compressed RFP workflows that previously took days down to minutes. Doctolib rolled the platform out to 3,000 employees and retired older intranet tools as a result.
The Series B brings Dust's total funding to over $60 million, with this round being the largest to date. Abstract and Sequoia led the investment. Snowflake Ventures and Datadog joined as participants.
"Most enterprise AI today is single-player: one person, one prompt, no compounding. Dust is building the multiplayer system, where agents and humans share context and work together across the entire company. Zero churn and 70% weekly active usage tell you this isn't experimental anymore. This is how enterprises will actually operate." — Konstantine Buhler, Partner at Sequoia
