Dust Overview
This Dust review covers one of the most interesting AI agent platforms for teams in 2026. Dust lets companies build custom AI assistants — called agents — that are grounded on your organization’s own data and can take real actions inside your tools. Instead of every employee pasting documents into a chatbot, Dust connects to Slack, Notion, Google Drive, GitHub, and 100+ other sources, keeping knowledge synced automatically.
What is Dust?
Dust is a Paris-based AI platform, founded in 2023, built for teams that want AI that actually knows their company. You create specialized assistants for different purposes — a support agent that answers from your help docs, an engineering agent that reads your codebase, a sales agent that pulls from your CRM — and each one draws on live company data rather than generic training knowledge. The core platform is open source under the MIT license, while the hosted service handles the infrastructure, security, and scale.
What sets Dust apart from a general chatbot is that its agents act. Beyond answering questions, they can update databases, create tasks, send emails, and trigger workflows across your connected tools. It is model-agnostic too: you can choose from 20+ models from OpenAI, Anthropic, Google, Mistral, and DeepSeek, picking the best model for each agent instead of being locked into one vendor.
How Dust Works
An admin connects the company’s data sources — Slack, Notion, Drive, GitHub, Salesforce, Zendesk, and more — and Dust indexes them with the same permissions your team already has, so people only see what they’re allowed to see. Then anyone can build an agent: give it a name, instructions, the data sources it may use, the actions it may take, and the model it runs on. Agents are reachable from a shared conversation interface, a Slack integration, a Chrome extension, a Raycast extension, and a CLI.
For developers, Dust exposes a REST API, webhooks, and Model Context Protocol (MCP) servers — both as a client and a server — so agents can plug into existing engineering workflows. On the governance side, it offers SOC 2 Type II compliance, GDPR and HIPAA-ready controls, audit logs, SSO, and a zero-data-retention option, which is why regulated and enterprise teams take it seriously.
Pricing and Credits
Dust moved to seat-based pricing in 2026. The Business plan mixes seat types: Free seats at $0 with a small lifetime credit allowance for trying things out, Pro seats at $30 per user per month ($24 on annual billing) with 8,000 credits monthly, and Max seats at $150 per user per month ($120 annual) with 40,000 credits monthly. Credits (measured in AWU, where 1 AWU equals $0.01) pay for AI usage depending on the model and task complexity. Enterprise plans use custom pricing with SSO, SCIM, larger storage, and dedicated support.
The credit model is fairer than flat per-message pricing because heavy users on powerful models pay proportionally more — but it does mean costs need monitoring. Admins can mix seat types so power users get Max while everyone else stays on Pro or Free.
Who Should Use Dust?
Dust fits companies of 10 to 1,000+ people that want AI embedded in how work gets done — support teams answering from live docs, engineers querying code context, marketers pulling from shared research. IT and security teams will appreciate the permission mirroring and compliance posture, which is often the blocker that keeps generic AI tools out of real companies.
It is a poor fit for individuals who just want a better chatbot, for very small teams where the per-seat cost outweighs the value, and for anyone who only needs occasional document Q&A — simpler tools cover that without the setup. Companies deeply embedded in Microsoft 365 may also prefer Copilot’s native integration over a third-party platform.
Our Verdict on Dust
Dust is one of the strongest team AI agent platforms available in 2026. The combination of live-synced company knowledge, model choice, real actions, and serious governance is hard to find elsewhere at this price point. The Paris team’s open-source core also means you’re not fully locked into a black box.
The bottom line of this Dust review: if your team keeps re-asking the same questions across Slack and Notion, Dust turns that scattered knowledge into agents that answer and act — just budget the per-seat cost carefully as you scale.
For a simpler individual-focused alternative, Lindy offers no-code AI employees for personal workflows without the team infrastructure. Browse more options in our AI Agents category.
Key Features
- Build custom AI assistants per team, role, or use case
- 100+ data source connectors with automatic syncing (Slack, Notion, GitHub, Drive)
- Model-agnostic: 20+ models from OpenAI, Anthropic, Google, Mistral, DeepSeek
- Agents take actions — update databases, send emails, trigger workflows
- MCP server and client support for connecting external tools
- REST API, webhooks, Slack app, Chrome and Raycast extensions
- SOC 2 Type II, GDPR, and HIPAA-ready security with zero-retention option
Dust Pricing
| Plan | Price |
|---|---|
| Free | $0 |
| Pro | $30/user/mo ($24 annual) |
| Max | $150/user/mo ($120 annual) |
| Enterprise | Custom |
Pricing checked on October 4, 2026 — always confirm on the official site.
Dust Pros & Cons
✓ Pros
- Agents grounded on live, auto-synced company data
- Model-agnostic — pick the best AI for each agent
- Agents take real actions, not just answer questions
- Strong governance: permissions mirroring, SOC 2, zero-retention option
- Open-source core avoids full vendor lock-in
✕ Cons
- No generous long-term free tier — free seats are limited to a small lifetime credit allowance
- Per-seat pricing of $30/user/month and up gets expensive quickly for large teams
- Overkill for individuals or anyone who only needs occasional one-off Q&A
- Designing good agents and connecting data sources takes real onboarding time
Dust FAQs
How much does Dust cost in 2026?
What are Dust credits (AWU)?
Can individuals use Dust, or is it only for companies?
How is Dust different from ChatGPT Team or Microsoft Copilot?
Is my company data safe with Dust?
Best Dust Alternatives
SmythOS
★★★★☆SmythOS reviewed for 2026 — the open-source agent OS with prompt-to-agent building. Honest look at features, pricing from $0, and who it's really for.
Langflow
★★★★☆Langflow reviewed for 2026 — the free, open-source visual builder for AI agents and RAG pipelines with 140k+ stars. Honest features, pros, cons, and verdict.
Review of AutoGen
★★★★☆AutoGen reviewed for 2026 — Microsoft's open-source multi-agent framework. Honest features, free pricing, maintenance-mode status, and verdict.
Manus
★★★★☆Manus reviewed for 2026 — the most autonomous AI agent yet, from $20/mo with credit billing. Features, real costs, honest pros and cons, and verdict.