Dust

Dust handles retrieval-augmented generation workflows such as document ingestion, chunking, embedding, and semantic search indexing. This reduces the need for teams to build and maintain their own vector databases or embedding infrastructure when creating AI agents grounded in company data.

The platform supports access to multiple foundation models from different providers, including GPT-4, Claude, Mistral, and Gemini according to the visible product description. Teams can configure model use for assistant needs, but exact routing behavior, model availability, and plan-level limits should be verified against Dust's current documentation.

Dust provides a graphical interface for designing multi-step LLM pipelines without writing code. Users can chain together retrieval steps, model calls, tool invocations, and data transformations into structured workflows. This makes it possible for product managers and operations teams to create and iterate on AI agent behavior while still offering depth for technical users.

Integrations with workplace tools, repositories, databases, APIs, and uploaded files can help Dust ingest and index organizational data for assistant use. Exact connector coverage, permission mapping, and synchronization behavior should be confirmed from Dust's current integration documentation.

Administrators can define which data sources and tools each AI agent can access, creating boundaries aligned with organizational security policies. Teams evaluating Dust for regulated or sensitive data should verify current compliance certifications, audit logging, SSO support, retention controls, and residency options directly with Dust.