Dust

Enables non-technical users to construct multi-step AI agents through a drag-and-drop interface without writing code. The builder abstracts tool orchestration, model selection, and data flow into visual blocks that chain together semantic search, API calls, and LLM reasoning steps. Agents are deployed immediately to a cloud runtime without compilation or deployment infrastructure.

Indexes documents from 10+ connected data sources (Google Drive, Notion, Confluence, GitHub, Slack, Zendesk, etc.) into a searchable knowledge base using semantic embeddings. Agents query this index with natural language to retrieve relevant context before generating responses, enabling RAG-style information retrieval without manual document management. Search results are ranked by semantic relevance and can be filtered by source or metadata.

Provides dashboards and metrics for monitoring agent performance (success rate, execution time, tool usage) and tracking costs (API calls, token consumption, model usage). Metrics are aggregated by agent, time period, and data source. Cost tracking shows spending by model provider and helps identify optimization opportunities.

Enables agents to navigate websites, fill forms, extract data from web pages, and interact with web applications programmatically. Agents can click buttons, type text, read page content, and follow links to complete multi-step web tasks. Web navigation is sandboxed and does not require agents to manage browser state or handle JavaScript rendering.

Enables agents to analyze structured data and query databases using natural language without requiring SQL knowledge. Agents can read data from Google Sheets, databases, and other structured sources, perform aggregations and transformations, and generate reports. Natural language is translated to queries internally, abstracting SQL complexity.

Dust enables teams to create and manage multiple versions of agents, test changes in staging environments, and deploy updates to production with rollback capabilities. Users can compare agent versions, track changes, and revert to previous versions if needed. The platform supports gradual rollouts (e.g., deploying to 10% of users first) and A/B testing different agent configurations.

Abstracts LLM provider differences by supporting GPT-5, Claude, Gemini, and Mistral models through a unified interface. Agents can be configured to use different models for different tasks, and the platform handles API key management, request routing, and error handling across providers. Model selection is configurable per agent or per step within an agent workflow.

Provides pre-built connectors to 10+ enterprise platforms (Slack, Google Drive, Notion, Confluence, GitHub, Zendesk, Salesforce, Chrome Extension) that handle authentication, data fetching, and schema mapping without custom code. Connectors support both read operations (querying data for agent context) and write operations (creating tickets, posting messages). Generic connectors (API, Google Sheets, Zapier) enable integration with any HTTP endpoint or workflow platform.

Agents execute within a supervised model where critical actions (e.g., sending messages, creating records, modifying data) can be configured to require human approval before execution. Execution logs show all tool invocations, model reasoning, and outputs, enabling humans to review and override agent decisions. Approval workflows are configurable per agent or per action type.

Enables enterprises to manage multiple agents across multiple workspaces with centralized governance controls. Features include user provisioning via SCIM (Enterprise tier), role-based access control, data residency options (US/EU), and advanced security controls. Workspaces are isolated environments where teams can build and deploy agents independently while maintaining organizational oversight.

Exposes agents as HTTP APIs that can be invoked programmatically from external applications, or as formulas in Google Sheets for non-technical users. API requests include task description and optional context, and responses include agent output and execution metadata. Spreadsheet integration enables users to invoke agents on rows of data without writing code, with results populated back into the sheet.

Captures detailed execution logs for every agent run, including tool invocations, model inputs/outputs, reasoning steps, and error messages. Logs are queryable and can be filtered by agent, date range, or status. Execution traces show the exact sequence of tool calls and model reasoning, enabling debugging of agent behavior and understanding of decision-making.

Provides pre-built agent templates for common enterprise use cases (customer support, sales, marketing, HR, legal, IT, engineering) that include pre-configured tools, prompts, and workflows. Templates serve as starting points that teams can customize without building agents from scratch. Each template is optimized for its domain with relevant data connectors and tool integrations.

Implements a zero-data-retention policy where Dust does not store user data, documents, or agent outputs after task completion. Data is processed in-memory during agent execution and discarded afterward. Organizations can configure data residency (US or EU) to comply with regional regulations. No data is used for model training or improvement without explicit opt-in.