Cross-functional AI programs
Organizations that want one governed layer for agents used across support, sales, marketing, engineering, analytics and internal knowledge.
Independent tool overview
Dust at a glance
Dust is a collaborative enterprise AI platform for building agents that can search company knowledge, use connected tools and work alongside people in shared workspaces. Its differentiator is not one proprietary model, but a governed layer across multiple models, data sources, tools, skills and team workflows.
Visit the official Dust site ↗
- Primary use
- Build and run collaborative AI agents on company context
- Model strategy
- Choose among 20+ frontier and open-source models
- Knowledge layer
- Live Connections plus files and shared context in Pods
- Action layer
- Native tools, MCP servers, API, webhooks and scheduled runs
- Published self-serve pricing
- Free, $30/month Pro and $150/month Max seats
- Enterprise security claims
- SOC 2 Type II, US/EU residency, SSO/SCIM and audit controls
Overview
What Dust is
Dust lets organizations create reusable agents with instructions, company knowledge and actions. Teams can choose among models from providers including OpenAI, Anthropic, Google, Mistral and others, then use those agents in Dust, Slack, Microsoft Teams, browser extensions, APIs and automated workflows.
Pods provide a multiplayer workspace where people and agents share conversations, files and tasks around a project. Dust also supports live Connections to systems such as Notion, Google Drive and Confluence, tools that can take actions in business apps, MCP servers, schedules, webhooks and cross-agent delegation.
The platform is best suited to companies willing to treat agent deployment as an operating program. Useful results depend on clean source data, narrow permissions, tested instructions, human approval for consequential actions and ongoing usage and quality monitoring—not simply connecting every available system.
Use cases
Who Dust is best for
The strongest fit depends on the job you need the product to complete, not the size of its feature list.
Knowledge-grounded agents
Teams that need assistants to search current company systems and cite internal context instead of operating as generic chatbots.
AI operators and builders
Technical or operations owners prepared to design skills, permissions, evaluations, rollout plans and reusable workflows for coworkers.
Capabilities
Core Dust features
Custom agents
Combines instructions, model choice, knowledge, skills and tools into reusable agents shared through a workspace.
Company data connections
Synchronizes supported sources such as Notion, Google Drive, Confluence, Slack, GitHub and business applications for retrieval.
Pods
Keeps people, agents, conversations, files and tasks together around a shared customer, project or initiative.
Tools and MCP
Allows configured agents to read from and take authorized actions in connected systems, including custom tools exposed through MCP.
Automation
Runs agents from schedules, webhooks and external automation platforms rather than requiring every task to begin with manual chat.
Governance and analytics
Provides access controls, agent management, audit logs, workspace analytics and usage or credit reporting at the applicable plan level.
Multiple model providers
Lets builders choose a model for each agent or workflow, with credit use varying by model and task complexity.
Process
How the Dust workflow works
Step 1
Choose one measurable workflow
Start with a repeated task that has a known owner, clear source of truth, measurable quality standard and safe fallback.
Step 2
Map data and action boundaries
List exactly what the agent needs to read and write. Create restricted spaces or Pods for sensitive work and avoid broad workspace-wide access by default.
Step 3
Build a narrow agent
Give it focused instructions, the smallest relevant knowledge set, an appropriate model and explicit rules for uncertainty and escalation.
Step 4
Test with a representative set
Evaluate correct answers, retrieval failures, fabricated claims, prompt injection, stale data, permission edges and attempted unsafe actions before rollout.
Step 5
Keep approval at the risk boundary
Require a person to approve external messages, customer-impacting changes, financial actions, deletions and sensitive-data decisions until reliability is proven.
Step 6
Roll out and observe
Train users, publish the agent's scope and monitor adoption, citations, errors, action logs, credit consumption and downstream business outcomes.
Step 7
Review continuously
Re-test when source permissions, models, tools or instructions change, and retire agents that have no owner or measurable value.
Cost
Dust pricing and free plan
Dust's Business plan allows a mix of Free, Pro and Max seats. Monthly list prices are $0, $30 and $150 per seat; annual billing reduces Pro to $24 and Max to $120 per seat per month. Enterprise uses custom pricing. Credits measure model and tool usage and reset each billing period for paid seats.
Free seat
$0
An occasional-use Business seat for evaluating Dust or light participation.
- 500 lifetime credits
- Free users are prompted to request an upgrade after credits are exhausted
- Can coexist with paid seat types in the same Business workspace
Pro seat
$30/month or $24/month billed yearly
The standard Business seat for most regular team members.
- 8,000 credits per seat per month
- Access to the Business platform's models, agents, knowledge and collaboration features
- Additional usage can depend on workspace overage settings
Max seat
$150/month or $120/month billed yearly
A high-usage Business seat for complex automations, deep research and tool-heavy work.
- 40,000 credits per seat per month
- Same model catalog; the larger allowance supports heavier usage
- Seat types can be mixed within a Business workspace
Enterprise
Custom
For larger deployments needing advanced governance, flexible usage and commercial support.
- Unlimited connectors and MCP servers as published on the pricing page
- Workspace-pooled credits and volume pricing
- Advanced security, administration, support and custom terms
- Confirm exact controls and deployment terms with Dust
Pricing checked . Check current pricing at the source ↗
Assessment
Dust strengths and limitations
Where it stands out
- Unifies agents, internal knowledge, business tools and human collaboration instead of treating each agent as an isolated chat.
- Offers model flexibility, reducing dependence on one model provider for every use case.
- Pods create persistent shared context for projects where work spans people, agents, files and tasks.
- Supports both live-synced knowledge and agent-written files, enabling read and write workflows.
- Publishes a self-serve seat and credit model while retaining an Enterprise option for larger deployments.
- The vendor publishes substantial security and governance controls, including SOC 2 Type II, SSO/SCIM, residency options and audit capabilities.
What to consider
- Dust requires implementation work: connectors, permissions, instructions, evaluations, owners and user training determine whether an agent is useful or risky.
- Credit consumption varies by model, reasoning depth and tool activity, so the nominal seat price does not by itself predict total cost for agent-heavy workflows.
- Agents can hallucinate, retrieve the wrong source or take an inappropriate action; Dust's own documentation recommends narrow sources and explicit uncertainty instructions.
- Connecting broad company systems can amplify existing access mistakes. Apply least privilege and test the effective permissions for each agent and user group.
- Everything inside a Pod is visible to all Pod members. Sensitive work belongs in a separate Restricted Pod with carefully reviewed membership.
- A Pod's agents can write files and perform tasks, making output governance and approval controls as important as read permissions.
- Security features and compliance claims may vary by plan, deployment and contract; buyers should review Dust's Trust Center, subprocessors, retention, incident terms and model-provider flow before approval.
- Model quality, availability and behavior can change independently because Dust relies on multiple third-party model providers.
- The platform can be excessive for an individual or small team that only needs general chat, basic document search or one simple automation.
Compare
Dust alternatives
The right alternative depends on the specific output, workflow, controls and budget your project requires.
Business Operations
Microsoft Copilot
A stronger fit for organizations centered on Microsoft 365 that prefer AI embedded directly in their existing productivity and identity stack.
Explore Microsoft Copilot →Project Management
Cassidy
Another company-context agent platform aimed at reusable assistants and automated workflows, with a different builder and integration experience.
Explore Cassidy →Business Operations
Lindy
A workflow-automation alternative for teams prioritizing no-code agents that execute repeatable tasks across business applications.
Explore Lindy →Questions
Dust FAQs
What is Dust AI?
Dust is a platform for teams to build and run AI agents with company knowledge, connected tools, shared workspaces, multiple model choices and governance controls.
How much does Dust cost?
As reviewed on August 31, 2026, Business seats are Free, Pro at $30 monthly or $24 monthly on annual billing, and Max at $150 monthly or $120 monthly on annual billing. Enterprise pricing is custom.
What are Dust credits?
Credits are Dust's unit for AI usage. Consumption depends on the model, task complexity and tools used. Free includes 500 lifetime credits, Pro includes 8,000 per month and Max includes 40,000 per month.
Can a company mix Free, Pro and Max seats?
Yes. Dust's pricing FAQ says these are seat types within the Business plan and administrators can assign them according to each user's expected usage.
Which AI models does Dust support?
Dust advertises more than 20 models from providers including OpenAI, Anthropic, Google, Mistral and DeepSeek. The builder chooses the model per agent, and more capable models may consume more credits.
What is a Dust Pod?
A Pod is a shared project workspace containing conversations, files and tasks for people and agents. All Pod members can see its contents, so access should match the sensitivity of the work.
Does Dust train models on company data?
Dust's current website states zero model training on customer data. Buyers should verify the complete data flow, model-provider terms, retention and contractual protections for their exact plan and deployment.
Can Dust agents take actions in other tools?
Yes, when the relevant native tool, connector, MCP server or API integration is configured. Treat write access as privileged and require approval for consequential actions.
Is Dust suitable for regulated or sensitive data?
Dust publishes enterprise security and compliance options, but suitability depends on the plan, deployment, data type and required agreements. Complete a legal, security and privacy review rather than relying only on marketing badges.
Bottom line
Our Dust verdict
Dust is a strong option for organizations that want AI agents to become shared infrastructure rather than a collection of private chats. Its mix of model choice, company context, tools, Pods and enterprise controls is compelling, while the Free/Pro/Max seat structure supports gradual rollout. The buying decision should hinge on operational readiness: without scoped permissions, evaluation sets, owners and approval boundaries, the same connectivity that makes Dust powerful can magnify mistakes.
Visit Dust website ↗Similar Business Operations Tools
View all Business Operations AI tools →
AI training for the future of work.
Get access to all our AI courses, hundreds of real-world AI use cases, live expert-led workshops, an exclusive network of AI early adopters, and more.
AI Courses
Get unlimited access to all of our current & upcoming industry-specific AI courses for the duration of your subscription.
Daily Guides
To keep up with the rapid pace of AI, our team publishes AI implementation guides daily. Our library contains 300+ practical use cases to automate real-world work.
Workshops
Join weekly, live, interactive sessions with industry leaders who are at the forefront of AI for hands-on implementation guidance and exclusive insights.
Community
Network with an exclusive community of AI-first professionals who are working smarter with AI. Learn how early adopters are using AI in their work and businesses.