Dust raises $40M to turn solo AI chatbots into team players
What's the deal? Dust, a Paris-based startup officially known as Permutation Labs SAS, has raised $40M in a Series B round to help enterprises move beyond isolated AI assistants toward collaborative, "multiplayer" agent ecosystems. The round was led by Abstract and Sequoia Capital, with participation from Snowflake and Datadog, bringing Dust's total funding to over $60M.
The company's platform transforms AI agents from personal chatbots into shared team resources that work across an entire organisation — with governance built in.
Why now? Most enterprises today are stuck in what Dust calls a "single-player" AI game. Employees use various chatbots and copilots that operate in silos: when a salesperson spends an hour researching an account with AI, a solutions engineer often repeats the same work the next day using their own tool. The context never compounds.
Traditional enterprise tools were designed for one-to-one interactions, so most AI systems don't create shared memory of the work they perform. Dust sees this as a massive inefficiency ripe for disruption.
What could go wrong? Shared AI workspaces raise thorny questions around data access and governance. When agents can pull from over 100 enterprise data platforms — including Slack, Notion, and Salesforce — ensuring the right people see the right information becomes critical. Enterprises with strict data compartmentalisation requirements may find the transition challenging.
There's also the question of whether organisations will actually change their workflows. Individual AI assistants are easy to adopt precisely because they're personal; getting teams to collaborate through shared agent surfaces demands cultural shifts, not just better tooling.
The signal: Dust's raise reflects a broader market shift from AI as a personal productivity tool to AI as organisational infrastructure. The involvement of Snowflake and Datadog — two major data platform companies — as investors signals that the enterprise data layer is increasingly seen as the foundation for agentic AI.
The startup has also developed "AI operators" — non-technical agentic roles in marketing, sales, and support that can build and deploy specialised agents without engineering help. Combined with integrated memory loops and a cloud-hosted compute environment, this points toward a future where AI agents aren't just assistants but persistent, learning teammates embedded across the business.
Read more: siliconangle.com
