Dust, the Paris-based AI collaboration platform founded three years ago by Samuel Biward and Stanislas Polu, has closed a $40 million Series B round led by Abstract, with participation from Sequoia, Snowflake Ventures, and Datadog. The funding brings the company's total raised to over $50 million as it pushes to build what it calls "multiplayer AI"—a system where humans and agents work together in parallel with shared context, tools, and aligned goals. Dust is already being used by more than 3,000 organizations globally that have collectively deployed over 300,000 agents across sales, product, operations, and customer success workflows.

The Single-Player AI Bottleneck

Dust's founders argue that most enterprises are trapped in what they call "single-player AI mode." Everyone gets their own agent that can handle individual tasks—researching a prospect before a call, drafting a presentation—but those gains stay siloed with each person. A sales rep might save time using AI to prep for a deal, but the solutions engineer running technical discovery the next day starts from scratch. The product marketer drafts one-pagers in isolation while the content writer builds blog posts from their own version of the brief. As organizations delegate more work to AI, Dust believes the real bottleneck shifts from generation to coordination: can teams coordinate across many people and many agents simultaneously with seamless reviews, approvals, handoffs, and shared visibility?

What Multiplayer AI Actually Means

Dust built its platform around persistent, shared workspaces where agents collaborate alongside humans rather than operating in isolated sessions. Every participant has access to the same context, tools, skills, and work already completed. The company uses a hybrid approach combining semantic search for deep contextual understanding with MCP (Model Context Protocol) connections for querying and acting across integrated tools like Slack, Snowflake, HubSpot, Notion, Gmail, and Google Drive.

Self-Improvement and Observability

Unlike static agents that perform exactly as initially configured until manually updated, Dust creates a continuous improvement loop where every interaction generates signal. Teams refine workflows over time, agents accumulate memory and context, and the system identifies patterns across conversations to suggest improvements to agent instructions and skills. For enterprise deployments, admins get granular permissions, cost and usage monitoring, full audit trails, and agent analytics—all in one place. The platform is SOC 2 Type II certified, GDPR compliant with EU/US data residency options, and guarantees zero model training on customer data.

Real-World Deployment at Scale

Dust's customers include Clay, where it serves as foundational knowledge infrastructure for a growing GTM team; Profound, which uses it as the source of truth for customer intelligence and post-sales operations; Persona, deploying more than 300 agents across 11 departments; and Doctolib, central to a company-wide AI strategy reaching 3,000 employees. The founders used their own platform to coordinate this announcement—running parallel workstreams with multiple people and agents contributing simultaneously while humans handled decisions, refined messaging, and coordinated approvals.

The AI Operators Thesis

Dust positions itself as built for what it calls "AI Operators": the growth marketers rebuilding outbound workflows, RevOps managers bridging gaps across the stack, GTM engineers rethinking how revenue gets generated, data analysts automating reporting pipelines, product marketers orchestrating complex launches, support managers automating escalations and renewals, and recruiters compressing hiring cycles. The company argues that these functional experts—not external consultants or centralized innovation teams—will be the ones who actually transform organizations with AI.

Key Takeaways

  • $40M Series B led by Abstract with Sequoia, Snowflake Ventures, and Datadog participation
  • Platform now serves 3,000+ organizations with 300,000 agents deployed globally
  • Core thesis: enterprise AI gains are bottlenecked by coordination, not capability
  • Multiplayer AI requires shared workspaces, hybrid knowledge layers, and continuous agent self-improvement

The Bottom Line

Dust's pitch cuts through the noise—individual agent productivity is table stakes now; the real unlock is getting humans and agents to compound work together. If their multiplayer vision delivers on even half of what they've outlined, this round positions them as foundational infrastructure for how AI-native organizations actually operate, not just another chatbot wrapper with enterprise branding.