The IT Nerd

Most companies have adopted AI, but they haven’t become meaningfully more intelligent as organizations. One person prompts an assistant, gets an answer, and the context disappears into a private chat window. The result is real productivity at the individual level, with very little compounding across teams. Dust, the multiplayer agentic AI system, was built to change that by making AI collaborative, shared, and operational across an entire company.

The company today announced a $40 million Series B with Abstract and Sequoia, with participation from Snowflake Ventures and Datadog. With this round, Dust has raised over $60 million in total funding. 

Why this matters now

Most organizations are stuck in what Dust calls single-player AI. Every employee has their own assistant with its own context and its own outputs. A sales rep researches an account, then the solutions engineer starts from scratch the next day. Marketing drafts a one-pager, then enablement recreates a battlecard with different inputs. The effort repeats, knowledge fragments, and gains don’t compound.

Dust argues that most AI tools used by enterprises reinforce this pattern. Foundation model workspaces and copilots are powerful, but they’re primarily designed around one individual’s workflows and context. Enterprise search tools retrieve information, but don’t take action. The outcome is more activity and more AI usage at the individual user-level, but not an intentionally designed system that compounds AI into shared leverage.

What Dust is building

Dust is the multiplayer AI system for human-agent collaboration. It gives business teams a platform to build, deploy, and manage AI agents that collaborate across an organization, connected to company knowledge, integrated with the tools teams already use, and governed with enterprise-grade controls.

At the center of Dust is a collaborative surface where people and agents work together across shared context, tools, conversations, tasks, and goals. Agents can analyze, transform, and generate files — including documents, spreadsheets, presentations, and interactive data visualizations — and take action across connected systems through Dust’s context layer, which combines semantic search across company knowledge with integrations to more than 100 data sources and business tools. Built-in memory and feedback loops help agents improve over time by learning from team preferences, usage patterns, and feedback, while proactively recommending improvements.

Dust is designed for enterprise deployment, with granular permissions, cost and usage monitoring, audit trails, and agent analytics. The platform is SOC 2 Type II certified, GDPR compliant, supports EU and US data residency, and does not train models on customer data, as contractually guaranteed by major model providers.

Dust runs primarily on its own product and is defining an emerging identity inside high-growth companies: AI Operators. These are the people closest to the work, inside functions like Ops, Support, Marketing, and Sales, who build and run AI systems for their teams, rewiring how work gets done from inside the business.

Traction and customer outcomes

Dust is used by more than 3,000 organizations globally, from high-growth AI-native companies to established enterprises. Monthly active adoption is consistently above 90%, with weekly active usage above 70% across customers, signaling that Dust has become embedded in how teams work. More than 300,000 agents have been deployed across the platform. In 2025, Dust saw significant customer expansion and acquisition, reaching 240% NRR with zero churn.

At Clay, Dust serves as foundational knowledge infrastructure for the rapidly growing GTM team, enabling the team to grow 4x without a proportional increase in enablement headcount. Profound uses Dust as the source of truth for customer intelligence and post-sales, compressing new hire ramp time from months to days. At Persona, teams across 11 departments have deployed over 300 Dust agents to condense cross-functional workflows like sales RFPs from days to minutes. Doctolib has made Dust central to its company-wide AI strategy, giving 3,000 employees smoother access to corporate information and enabling the decommissioning of legacy intranet tools. 

The origin 

Dust was founded by Gabriel Hubert and Stanislas Polu, who have been building together since meeting at Stanford in 2007. They previously co-founded TOTEMS, a data analytics company acquired by Stripe in 2014, and spent five years at Stripe scaling products and teams. Polu later joined OpenAI as a research engineer on Greg Brockman’s team, co-authoring papers on AI reasoning with Ilya Sutskever. Hubert became Chief Product Officer at Alan.

In September 2022, Polu left OpenAI with a conviction that became Dust’s founding thesis: the models were already powerful enough to be economically transformative, but were under-deployed because the product layer was missing. Dust incorporated in February 2023 to build that horizontal layer on top of frontier models and company knowledge, with a model-agnostic approach that avoids vendor lock-in.

What’s next

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. The bet is that the next phase of enterprise AI won’t be won by who has the best single assistant. It’ll be won by who turns AI into shared, compounding capability across the entire org.