Dust

Dust Review

Dust is a platform for building AI assistants that connect to company knowledge and tools. Teams define playbooks that combine retrieval with function calls—search docs, query a database, create tickets—then publish assistants to Slack, the web, or custom apps. Developers manage prompts, evaluation datasets, and observability to track drift and quality, while permissions keep outputs within access scopes. Integrations include connectors for drives, wikis, CRMs, and analytics sources. Typical workflows include support copilot, sales research, and ops automation. The value is reliable, auditable assistants that actually act, not just chat.

Things to Know About Dust

Dust drawbacks: Requires strong data hygiene for reliable retrieval and agent behavior; hallucinations occur without citations or guardrails. Enterprise governance—SSO, audit, data residency—may need custom setup, and integration depth varies by system. Observability of prompts/tools is improving but can be opaque for production incident response.

Top Features

• Workspace to build AI assistants that read company docs and tools
• Connectors for Google Drive, Notion, Slack, GitHub, and email
• Retrieval-augmented answers with citations and attachments
• No-code flows, tools/functions, and guardrails per assistant
• Granular permissions mirroring source access controls
• Conversation memory, shareable threads, and handoffs
• Observability: logs, prompts, evaluations, and analytics
• Versioning, staging, and approval workflows for changes
• API and webhooks for embedding into internal apps
• Compliance features: SSO/SCIM, data residency, and redaction

Dust Pricing

Dust pricing: quote-based pricing aligned to users, model routing, and data governance; business tiers add workspaces, permissions, and analytics; enterprise plans include SSO, audit logs, and SLAs; pilots are common before larger deployments.

How to use Dust

To use Dust, create a workspace, connect data sources, and design an assistant by setting instructions, tools, and retrieval. Test with representative queries, enforce redaction and guardrails, and deploy via web or API. Monitor logs, collect feedback, and retrain or adjust prompts to improve accuracy.

Alternatives & Competitors

Dust competes with LangChain/LLM orchestration tools, Retool/Make for workflows, and enterprise assistants like Microsoft Copilot Studio—platforms to build secure, retrieval-augmented AI apps. Overlap includes connectors, prompt/version management, evals, and role-based access. Rivals may add vector DB choices, fine-tuning, on-prem deployment, and audit/compliance exports. Dust’s strengths are fast composition of RAG agents and team sharing. Gaps can include breadth of enterprise integrations, deep MLOps pipelines, and regulated-industry controls found in larger platforms.

Video