Dust

What is Dust?

Dust is an AI agent workspace designed for teams and organizations rather than individual chatbot use. Its main purpose is to let people create specialized agents that can search approved company information, work with connected applications, analyze data, produce documents, and carry out defined actions.

A typical Dust agent combines custom instructions with knowledge sources, tools, skills, optional memory, and a selected language model. This makes Dust closer to an agent-building and workflow platform than to a general-purpose conversational assistant. It can support everyday questions, but its value is greatest when an organization wants repeatable AI workflows connected to its existing systems.

How teams use Dust

Users generally begin with an existing agent or create one for a specific responsibility, such as internal research, customer support, sales operations, reporting, or engineering assistance. They can provide text, documents, images, URLs, structured data, or information retrieved from connected applications. The agent then searches authorized sources, uses configured tools, and returns an answer or performs an action.

Dust organizes this work in shared Pods, which can contain conversations, files, tasks, people, and agents around a project or initiative. This gives teams a place to review results and coordinate human and agent work instead of keeping every interaction in a private chat.

Knowledge-connected agents

One of Dust’s central capabilities is connecting agents to company knowledge. Integrations can make information from services such as Notion, Google Drive, SharePoint, OneDrive, Slack, GitHub, project-management systems, and business applications available to authorized agents. The exact connectors and permissions depend on the workspace and plan.

This allows an agent to answer questions about internal procedures, summarize documents or conversations, prepare reports from organizational information, and retrieve context before taking an action. The quality of the result still depends on the quality, freshness, permissions, and organization of the connected data; Dust does not eliminate those underlying knowledge-management issues.

Workflow automation and integrations

Dust supports multi-step workflows in which agents can be chained together or combined with application actions. Scheduled runs can produce recurring briefings or reports, while event triggers and webhooks can start work when something happens in an external system. Configured agents can also create tasks, update connected services, and coordinate work across applications.

The platform documents more than 70 integrations and also supports custom or remote MCP servers. This is important for teams that need agents to operate within existing business processes rather than simply generate text. Developer APIs and embeddable agent experiences provide additional ways to expose Dust-powered agents inside other products or workflows.

Models and agent configuration

Dust lets workspaces select among documented models from providers including OpenAI, Anthropic, Mistral AI, and DeepSeek, as well as Google and other supported providers. The available model list can change over time and may vary by workspace.

Model selection matters because different tasks can have different requirements for reasoning, speed, cost, retrieval, coding, or structured output. Dust’s configuration layer lets an organization define which model and tools an agent should use, while its credit system accounts for usage across models and actions.

Important capabilities

• Custom agents: Create agents with specialized instructions, knowledge sources, skills, tools, and optional memory.
• Company knowledge retrieval: Search authorized information from connected business applications and internal sources.
• Workflow orchestration: Chain agents, schedule recurring work, respond to events, and connect actions through webhooks or application integrations.
• Data and document work: Analyze connected data, summarize material, create structured documents, and use code execution where configured.
• Collaborative Pods: Keep agents, conversations, tasks, files, and teammates together around shared work.
• Developer access: Use APIs, embedding options, MCP connections, and webhooks to integrate agents into broader systems.

Who is Dust for?

Dust is primarily suited to organizations that need governed AI agents connected to internal knowledge and business systems. Likely users include operations teams, support and sales groups, researchers, analysts, engineers, and cross-functional teams that repeatedly perform information-heavy tasks.

It can be a good fit when an organization wants more control and structure than a general chatbot provides. It is also relevant for teams comparing visual agent-building platforms such as Botpress, Dify, or Langflow, although the best choice depends on the desired balance between collaboration, enterprise governance, integrations, development control, and workflow design.

Dust is less suitable for someone who only wants a simple personal chatbot, a standalone writing application, or a creative image, video, or audio generator. Its configuration and integration features can be unnecessary overhead for lightweight individual use.

Pricing and access

Dust has a continuing free Business seat with 500 lifetime credits. Paid Pro seats start at $24 per seat per month when billed annually, or $30 per seat per month with monthly billing. Max seats are priced higher and include a larger credit allocation. Enterprise pricing is custom.

Usage is credit-based rather than being defined only by the number of conversations. Pro seats include 8,000 credits per seat per month, while Max seats include 40,000 credits per seat per month. Credit consumption varies according to the model, task complexity, retrieval, research, code execution, and connected-tool actions. Unused monthly credits do not roll over. A 14-day trial of the paid experience may be offered, while the free seat remains a separate access option.

Privacy, security, and administration

Dust is intended for organizational data and provides enterprise-oriented controls such as encryption at rest and in transit, role-based access, private spaces, SSO, US or EU data residency, and SOC 2 Type II certification. Enterprise features include options such as audit logs, SCIM provisioning, custom retention, and single-tenant deployment.

Dust states that customer data is not used to train models and that its third-party model providers operate with zero data retention for Dust data. These statements do not mean that every workspace has identical retention or governance settings. Platform retention depends on the plan and configuration, so organizations should review the applicable policy and configure connected sources carefully.

Limitations to consider

• More setup than a chatbot: Useful results often require careful agent instructions, source selection, permissions, and workflow design.
• Usage-based cost: Seat prices do not tell the whole cost story because credit consumption varies by model and task.
• Plan-dependent governance: Some advanced security, analytics, connector, retention, and deployment features require higher-tier plans.
• Changing availability: Models, integrations, and features can change over time and may differ between workspaces.
• Not a media-production tool: Dust’s main focus is knowledge work, agents, documents, data, and connected actions rather than image, video, music, or voice generation.

Is Dust a good fit?

Dust is a strong fit for teams that want to build reusable agents around internal information and existing work systems. It is particularly useful when the goal is to automate recurring research, reporting, support, operations, or coordination tasks while retaining workspace-level controls.

Individuals looking for instant answers with minimal configuration may find it more complex than necessary. Organizations should also assess the quality of their source data, connector permissions, expected credit usage, and required enterprise controls before adopting it. Dust is best understood as a collaborative agent and workflow layer for business operations, not simply as another general-purpose AI chat interface.