Dust

Dust is an open-source, multiplayer AI agent platform built by dust-tt that lets teams and AI agents collaborate as co-contributors in a shared workspace. The core repository is written primarily in TypeScript, licensed under MIT, and hosted on GitHub with over 1,300 stars. The commercial product at dust.tt targets enterprise teams and positions itself as a platform for "AI Operators" — people who build and run AI workflows for their entire organization.

What It Is

Dust is a custom AI agent platform that connects to a company's existing tools and data sources, giving both human teammates and AI agents shared access to the same knowledge, conversations, and notifications. Rather than single-player AI usage, Dust is designed around a "multiplayer" model where agents and people work side by side. The platform supports model flexibility, letting teams switch between frontier models from OpenAI, Anthropic, Google, and Mistral.

How the Multiplayer Workspace Works

Dust's core concept is a shared collaboration surface where agents are treated as equal co-contributors alongside human teammates. Key architectural ideas include:

• Semantic knowledge layer: Agents don't just retrieve documents — the platform synthesizes company knowledge so agents can understand context, not just search it.
• Self-improving agents: Best practices consolidate into shared skills that spread automatically across the workspace.
• Dual-layer permission model: Separates what agents can access from who can use them, with SCIM-synced groups and admin-gated overrides.
• 100+ production connectors: Bi-directional sync with read and write support, incremental data refresh, and webhook support for event-driven workflows.

Team Use Cases

Dust is designed to serve multiple departments from a single platform:

• Engineering: Code debugging with IDE context, automated code reviews, incident response runbooks, and continuous doc generation.
• Customer Support: Ticket triage and routing, draft responses grounded in the knowledge base, escalation detection.
• Sales: Account research briefs, lead qualification, proposal drafting, CRM hygiene automation.
• Marketing & Content: Brief generation, content localization, performance analysis, social drafting.
• Data & Analytics: Self-service analytics in plain English, pipeline monitoring, metrics standardization.

Enterprise Security and Scale

Dust's security posture is documented in a trust datasheet with 19 listed controls. Vendor-published claims include SOC 2 Type II certification, GDPR compliance with EU data residency, HIPAA-ready deployment, SSO (SAML, OIDC) + SCIM, 365-day audit log retention, AES-256 encryption at rest, TLS 1.3 in transit, and a stated 99.9% uptime SLA. The platform claims support for 10,000+ users per workspace and sub-2s response time at p95.

Update: Series B and Recent Growth

In May 2026, Dust announced a $40M Series B round, as reported by Axios and Forbes. The company's homepage states it has 300,000+ agents deployed across 3,000+ teams. The latest GitHub release is dsbx-v0.1.14, published on May 21, 2026, reflecting active development. The repository was last pushed to on May 21, 2026, signaling ongoing product movement. The Series B announcement was framed around "multiplayer AI" as the company's next growth chapter.

Open-Source Lineage and Developer Platform

The core Dust platform is open source under the MIT License on GitHub (dust-tt/dust), created in August 2022. Dust also offers a developer platform with a RESTful API, MCP (Model Context Protocol) support for proprietary systems, OAuth2 for third-party permissions, and platform documentation at docs.dust.tt. The GitHub repository has 263 forks and 132 open issues as of the latest data.