Most workplace AI has a critical flaw: it disappears the moment it’s used.
A sales rep researches a customer account, painstakingly refining their prompt to get something genuinely useful. They get what they need, then close the tab.
The next day, their colleague opens a new chat and starts the whole process from scratch, investing time and energy into repeating work that’s already been done.
That hard work never reaches marketing, or product. Institutional knowledge stays stubbornly siloed: locked in private chat windows, lost in closed tabs, reset with every new conversation.
Dust calls this pattern “single-player AI”: the individual gains are real – but they don’t trickle down to the wider organisation.
This is the problem Dust was built to solve. Today, we’re proud to participate in their $40m Series B led by Abstract and Sequoia, with participation from Snowflake Ventures, Datadog, Wing, Aglaé, and XYZ. This round brings Dust’s total funding to over $60m.
The Dust thesis is that the next leap in enterprise AI won’t come from a better model or a faster assistant. It’ll come from a fundamentally different kind of system: one where humans and agents seamlessly share context, work in the same workspace, and compound each other’s output over time. They call this multiplayer AI.
The platform Dust has built reflects that thesis. Teams use it to build, deploy, and manage agents connected to company knowledge and integrated into the tools they already use. Rather than each employee having a private assistant, the whole organisation works from a shared collaboration surface. Agents learn from every interaction, improving over time rather than constantly starting from scratch.
Founders Gabriel Hubert and Stanislas Polu have been building together since meeting at Stanford in 2007. They co-founded TOTEMS, a marketing tools and data analytics company which we invested in back in 2011, and which was eventually acquired by Stripe in 2014. Similarly, TOTEMs was built on the belief that teams make better decisions with the right data and information in front of them. Dust takes that same principle and applies it to AI.
Polu went on to join OpenAI as a research engineer, co-authoring papers on reasoning with Ilya Sutskever. Hubert became CPO at Alan. In 2022, Polu left, driven by a specific conviction: that AI models were already powerful enough to be economically transformative, but were under-deployed because the product layer was missing. Along with Hubert, he launched Dust in February 2023 to build that layer.
The numbers speak for themselves: Dust is used by more than 3,000 organizations globally, from fast-moving startups like Clay and Vanta to large enterprises like Doctolib.
Monthly active adoption is consistently above 90%, with weekly active usage above 70% across customers, signaling how embedded Dust has become in how teams work. Since launch, more than 300,000 agents have been deployed across the platform. This raise follows a period of significant customer expansion and acquisition for Dust, reaching 240% NRR with zero churn in 2025.
“This is a century-defining transformation, and we’re only in year three,” said Gabriel Hubert. “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.”
"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.”Gabriel Hubert ~ Dust
Dust plans to use this round to push three frontiers at once: agents that learn and improve automatically as they’re used, collaboration primitives that make humans and agents equal co-contributors with bidirectional access to shared projects, tools, and context, and infrastructure that makes governance and orchestration predictable at enterprise scale.
Seedcamp Partner Carlos Eduardo Espinal welcomed the news:
“We first backed Gabriel and Stan in 2011, when they were building TOTEMS, long before AI was mainstream.”
“Even then, they were deeply engaged with how teams can use shared context and information to make better decisions. Dust is the fullest expression of that conviction – and with 70% active usage across 3000 organisations globally, it’s clear the market agrees.”
“The shift from individual AI tools to shared, compounding systems is one of the most important transitions happening in enterprise technology right now, and Dust is at the forefront.”