Dust AI Review 2026: Team Agents, Pricing, Security, and Fit

Short answer

Dust is worth evaluating for cross-functional teams that want shared AI agents connected to company knowledge and tools while retaining model choice and governance. Its self-serve Pro tier gives smaller teams a credible starting point. The decisive issue is not how many agents you can build; it is whether permissions, retrieval, actions, audit logs, human review, and credit economics remain predictable on real work.

Best for

• Cross-functional teams building shared agents
• Companies needing model flexibility
• Teams prepared to govern connectors and permissions

Look elsewhere if

• Unbounded autonomous actions
• Buyers unable to forecast usage credits
• Sensitive deployments without security review

What Dust verifiably does

Dust documents custom and global agents, shared workspaces, reusable skills, schedules and triggers, more than 70 connectors, more than 20 frontier and open-source models, MCP, collaboration, permissions, analytics, and enterprise deployment options. Agents can retrieve company context and use tools for search, analysis, code execution, and actions in connected applications.

Important limitations

Credits are workload-dependent, so seat price alone does not predict total cost. Connecting Slack, Drive, GitHub, Salesforce, Zendesk, or other systems expands the accessible data surface and may sync more than a user expects. Retrieval can miss or mis-rank sources, agents can take incorrect actions, and retention, audit logs, SCIM, single tenancy, and some residency controls are enterprise features.

Pricing snapshot

Free includes 500 lifetime credits. Pro is $24 per seat/month with 8,000 credits per seat; Max is $120 with 40,000 credits per seat. Enterprise uses pooled credits, volume pricing, and custom terms. Credit use varies by model, task complexity, retrieval, code execution, search, and connected actions. Reviewed August 24, 2026.

A fair buyer test

Give one cross-functional team two bounded agents: a read-only account brief and an approval-gated CRM hygiene workflow. Seed conflicting sources, revoked access, private records, and tool failures. Measure accepted answers, citation quality, blocked access, incorrect actions, escalations, latency, per-task credits, audit detail, and maintenance.

Final verdict

Dust earns a shortlist for teams that need collaborative agents, broad connectors, and model flexibility with a self-serve entry point. It is less attractive when workloads are hard to bound or the buyer cannot rigorously map connector scope, permissions, retention, and credit use.

This is a research-based assessment, not a claim of hands-on product testing. Product, pricing, privacy, security, and usage claims were checked against the first-party sources below on August 24, 2026. Verify current terms and run the proposed test with approved data before adoption.