Dust Raises $40M Series B to Push Enterprise AI Beyond Solo Chatbots

Dust just raised $40M in a Series B round led by Abstract and Sequoia, with participation from Snowflake and Datadog. The Paris and San Francisco-based startup is building what it calls “multiplayer AI” for the enterprise: shared AI agents connected to company knowledge, workflows, and operating systems instead of isolated chatbot experiences.

The funding arrives at a strange moment in enterprise software. Nearly every company now claims to have an “AI strategy,” yet most internal deployments still resemble expensive autocomplete wrapped inside Slack notifications. Enterprises bought copilots. What they actually needed was coordination. Dust is betting the next phase of enterprise AI will not revolve around smarter prompts but around shared context.

Dust founders Gabriel Hubert and Stanislas Polu understand that problem unusually well. Between Stripe, OpenAI, and Alan, the pair spent years inside systems where operational scale matters more than demo theater. That experience shows up in the product philosophy. Dust is less interested in replacing workers than in reducing the organizational drag that quietly eats entire workweeks alive.

What Happened

Dust announced a $40M Series B to accelerate development of its enterprise AI platform. The round was co-led by Abstract and Sequoia, with strategic participation from Snowflake and Datadog, two companies that understand infrastructure economics better than most of the AI market currently pretending to.

The company focuses on deploying AI agents across teams instead of limiting AI usage to isolated employee interactions. Dust connects models, internal documents, communication systems, SaaS applications, and workflows into shared workspaces where employees and AI agents operate together. That distinction matters because most enterprise AI products today function like extremely confident interns. Ask a question, receive an answer, move on. Dust is trying to build organizational memory instead.

Shared agents, persistent context, and cross-functional workflows sit at the center of Dust’s strategy. The company wants AI systems that understand not just what 1 employee asked, but how an organization actually works when deadlines, approvals, sales targets, compliance reviews, and customer escalations start colliding at 2:17 PM on a Wednesday. Dust says organizations have deployed more than 300,000 agents through the platform, while also reporting zero customer churn during 2025, a metric enterprise operators notice immediately because enterprise software customers abandon innovation experiments fast when adoption collapses after the executive demo.

Why Dust Matters in Enterprise AI

Enterprise AI adoption has entered an awkward adolescence. Boards want AI strategies. Executives want productivity gains. Employees mostly want systems that stop wasting their time. That gap created one of the biggest market distortions in software right now.

For the past 18 months, the industry flooded the market with single-user AI copilots. Helpful sometimes. Transformational rarely. Most deployments improved fragments of workflows while leaving organizational fragmentation intact. Companies added AI on top of operational chaos instead of reducing the chaos itself, and Dust’s positioning attacks that exact problem.

The company frames its platform as “multiplayer AI,” a phrase that sounds almost playful until you understand the strategic implication underneath it. Enterprises are collaborative systems. Knowledge lives across teams, systems, permissions, conversations, and historical decisions. AI becomes exponentially more valuable once context becomes shared infrastructure instead of private chat history.

That explains why investors like Sequoia, Snowflake, and Datadog are paying attention. The market is slowly realizing enterprise AI will not be won by whichever model writes the funniest LinkedIn post or summarizes meeting notes the fastest. The winners will control orchestration layers, workflow memory, governance systems, and organizational context. In other words, the plumbing. Nobody brags about plumbing until the building floods.

The Founders Built Dust From Operational Reality

Gabriel Hubert and Stanislas Polu are not first-time founders discovering enterprise software through Twitter discourse and venture capital podcasts. Their background matters because Dust’s product architecture reflects operators who have already lived through scaling pain.

Before Dust, Hubert and Polu co-founded TOTEMS, a data analytics startup later acquired by Stripe. Hubert later became a product leader at Alan, while Polu joined OpenAI as a research engineer focused on reasoning systems. That combination matters more than flashy branding.

Stripe teaches operational rigor. OpenAI teaches frontier model capability. Alan teaches workflow design at scale. Dust sits directly at the intersection of those disciplines. The platform feels less like AI as entertainment and more like infrastructure designed by people who understand how enterprises actually break internally, usually somewhere between Slack, spreadsheets, undocumented processes, and the sentence “Can someone resend the latest version?”

The enterprise stack has become a digital landfill of institutional memory. Dust is trying to turn that landfill into operational intelligence.

Enterprise AI Is Moving Toward Shared Systems

Dust’s rise reflects a broader transition happening across enterprise AI infrastructure. The first phase of generative AI focused on individual productivity: chatbots, writing assistants, summarization tools, fast dopamine, strong demos, and massive engagement spikes followed by uncertain retention curves.

The second phase is starting to look very different. Enterprises now care about integration depth, governance, permissions, auditability, reliability, and workflow continuity. They want AI systems that fit into organizational operations without creating security nightmares or productivity fragmentation. That shift benefits companies like Dust.

The startup integrates with tools like Slack, Notion, GitHub, and Google Drive while emphasizing shared organizational usage instead of isolated prompting. The company also highlights SOC 2 Type II certification and GDPR compliance, which may sound painfully unsexy until legal, security, and procurement teams enter the conversation and suddenly become the most powerful people in the buying cycle.

Enterprise software eventually becomes a trust business every single time.

The Competitive Landscape Is About Context, Not Models

Dust operates inside one of the most crowded categories in technology right now. Microsoft, OpenAI, Anthropic, Google, Atlassian, Notion, Slack, and dozens of AI-native startups are all chasing enterprise AI dominance. But model access alone is rapidly commoditizing.

The strategic battle is shifting toward orchestration, integration, and workflow ownership. The companies that win enterprise AI may not own the smartest models. They may own the systems enterprises trust to coordinate work safely across teams. That distinction changes valuation logic entirely.

Infrastructure layers historically capture enormous long-term value because replacing them becomes organizationally painful. Once AI agents become embedded inside operations, switching costs rise quickly. Governance, memory, integrations, and workflow dependency create stickiness far stronger than novelty features ever could.

That is likely part of the reason this funding round attracted infrastructure-heavy investors like Snowflake and Datadog. They understand where software gravity usually settles.

What This Signals for the Market

Dust’s funding round signals that enterprise AI is maturing beyond chatbot experimentation and moving toward organizational operating systems. The market no longer rewards AI products simply for existing. Enterprises increasingly want measurable operational outcomes, durable adoption, governance controls, and workflow integration.

The era of “look what the model can do” is slowly giving way to “show me how this changes operations without creating 5 new problems.” That transition will reshape the next generation of enterprise software winners.

AI agents are becoming less like standalone tools and more like coworkers embedded into systems, processes, and communication layers. The startups that survive this shift will likely be the ones building coordination infrastructure instead of novelty interfaces.

Dust appears to understand that better than most, not because the company talks louder than competitors, but because the product thesis sounds like it came from people who have actually sat inside enterprise complexity long enough to know where the bodies are buried.