Move your team from Glean to Dust.
Each person gets a short interview, then a personalized step-by-step guide showing exactly where their saved work, prompts, and projects land in Dust — and the few things that need a manual step, walked through. You send one invite; nobody becomes the help desk, and nothing in Glean is changed or cancelled. Want to see where you’re paying for both first? Run the free audit.
Dust agents invoke a tool library (web search, data-source search, table query, file generation, image generation, Slack/Notion/GitHub/Confluence/HubSpot/Salesforce/Gmail/Google Calendar/Outlook/Zendesk actions, MCP servers) that is configured per agent in its builder. Glean Agent Builder exposes a native actions library across the same major apps (Slack, Microsoft 365, Salesforce, Jira, GitHub, Google Workspace) plus any custom action importable via OpenAPI spec, arranged as step-level building blocks within the agent workflow. To migrate tool-heavy Dust agents to Glean: list every tool enabled on the Dust agent; find the matching Glean native action or import via OpenAPI; set up the same app authentication in Glean's connector admin; rebuild the invocation logic as Agent Builder steps. To migrate Glean action-steps to Dust: map each Glean step to a Dust tool (e.g. Glean 'Slack: send message' → Dust Slack action; Glean custom OpenAPI step → Dust MCP server); note that Dust tools are set at the agent level and invoked via RAG-guided instructions rather than explicit step order.
- Warning: Dust tool invocation is instruction-driven (the agent decides when and whether to call a tool based on natural-language instructions); Glean agent steps execute in a defined order with explicit branching — logic-heavy workflows behave differently when ported.
- Warning: Dust supports MCP servers for custom integrations; Glean supports custom actions via OpenAPI spec import — the configuration format differs and requires re-authoring the integration definition.
- Warning: Dust's 'Run Agent' delegation tool (chaining specialised agents) has no direct Glean equivalent; complex Dust multi-agent pipelines need to be restructured as single Glean agents with multiple steps or separate agents routed from Glean Assistant.
- Warning: Glean enforces tool-call budgets and payload caps per agent response; Dust does not publish equivalent hard limits, though programmatic credit consumption applies.
Both Dust and Glean provide an AI assistant that answers questions grounded in the company's connected data, with source citations, and accessible to all employees. In Dust, teams build and share a general-purpose assistant agent (or use the pre-built @dust agent) backed by Company Data spaces over connected sources; users invoke it via the web app, Slack, or browser extension. In Glean, Glean Assistant is the out-of-the-box conversational layer — employees query it at app.glean.com, via desktop spotlight, mobile, or browser extension, and it synthesises answers from indexed connectors with inline citations. To migrate from Dust to Glean: ensure all data sources connected to Dust's general assistant are also connected as Glean connectors; brief employees that Glean Assistant replaces their @dust habit with the Glean chat surface; note that Glean Assistant auto-routes complex requests to specialist Glean Agents. To migrate from Glean to Dust: build a Dust agent with broad Company Data access, write instructions covering the general assistant scope, publish it to the workspace, and connect the Dust Slack integration in place of Glean's Slack bot.
- Warning: Glean Assistant has fast-mode (unlimited) and extended-thinking/deep-research modes (FlexCredits); Dust charges programmatic credits for API/automation usage but human-typed messages are not credit-gated on active plans.
- Warning: Glean's underlying LLM is configurable at the admin level via Model Hub (OpenAI, Anthropic Claude, Google Gemini, Amazon Nova); Dust exposes per-agent model selection to builders, giving teams finer-grained control over model choice.
- Warning: Glean Assistant's Enterprise Graph enables multi-hop relational queries ('who owns feature X?') powered by a continuously updated knowledge graph; Dust's RAG is search-based over flat data sources and does not natively build entity relationship graphs.
Both Dust and Glean offer no-code builders where teams create custom AI agents grounded in company knowledge. In Dust, admins or team members open the Agent Builder, write natural-language instructions, attach data sources (Spaces, Company Data), select a model, and optionally add tools. In Glean, admins open Agent Builder, describe the agent goal (auto-generates steps) or build manually with drag-and-drop triggers, actions, and flow-control blocks. To migrate from Dust to Glean: export your Dust agent instruction text and replicate it in Glean's system-prompt/instructions field; re-map Dust data-source attachments to equivalent Glean connectors (both support Google Drive, Notion, Confluence, Slack, Jira, Salesforce, GitHub, Zendesk); re-create any Dust templates as Glean Agent Builder starting configurations. To migrate from Glean to Dust: export step definitions and map Glean 'actions' to Dust tools (web search, Slack, Notion, GitHub actions); recreate Glean branching/looping logic as separate specialised Dust agents chained via Run Agent delegation.
- Warning: Glean Agent Builder supports branching, looping, and multi-step flow control natively; Dust achieves complex multi-step flows by chaining agents via the 'Run Agent' tool, which requires a different design pattern.
- Warning: Dust lets builders choose the underlying LLM per agent (Claude, GPT-4, Gemini, Mistral, etc.) at creation time; Glean routes through its Model Hub controlled at the admin/workspace level, so individual agent creators do not freely swap models.
- Warning: Glean's Agent Builder includes a 30-version history and preview/test environment; Dust has no equivalent published version-history UI — rollback means manual re-editing.
- Warning: Dust Pro plan caps programmatic (API) credits at roughly $50/user/month; Glean advanced features (deep research, premium-model thinking) consume FlexCredits under Enterprise Flex, adding variable cost per agent run.
Both Dust and Glean provide the daily employee surface where staff access AI agents and company knowledge without leaving their normal workflow. Dust surfaces agents through: the native web app, a Slack integration (chat in channels or DMs), a browser extension (Chrome, Firefox, Chromium), and email forwarding. Glean's Work Hub is the personalised employee homepage with announcements, people directory, resources, calendar, and a personalised recommendations feed — also available via web, desktop app, mobile (iOS/Android), and a browser-extension new-tab/sidebar. To migrate from Dust to Glean: redirect users from the Dust web app to app.glean.com or the Glean browser extension; replace the Dust Slack integration with Glean's Slack connector and Glean Assistant Slack bot; point company links to Work Hub resource pinning instead of Dust Spaces. To migrate from Glean to Dust: install the Dust browser extension to replace Glean's browser-extension sidebar; connect the Dust Slack integration to replicate in-Slack AI access; note that Dust has no equivalent of Work Hub's announcements or people-directory features.
- Warning: Glean Work Hub includes announcements targeting by department/location, a people directory, and a personalised content feed — Dust has no equivalent homepage or people-directory features; teams relying on these must find separate tools.
- Warning: Dust supports email forwarding as an agent access surface; Glean has no equivalent email-based invocation path.
- Warning: Glean's desktop app (macOS spotlight-style launcher) and mobile apps (iOS/Android) have no counterpart in Dust, which is web-and-extension only.
- Warning: Dust programmatic/API access is billed separately from seat licences with monthly credit caps; Glean embeds programmatic API deployment within the Enterprise plan (advanced agent features may still consume FlexCredits).
Both Dust and Glean ingest enterprise app data via managed connectors so agents and search can operate over live company knowledge. In Dust, workspace admins go to Connections, add a connector (Google Drive, Notion, Confluence, GitHub, Slack, Intercom, Snowflake, BigQuery, Zendesk, Gong, Microsoft OneDrive/SharePoint/Teams, Outlook, Excel, plus beta: Jira, Salesforce, HubSpot, Guru, Linear, Dropbox, Front), and set folder/channel-level permission scopes. Syncs are automatic and near-real-time. In Glean, admins configure connectors from the 100+ catalogue (same major apps plus HR systems like Workday and BambooHR, design tools like Miro and Canva, and engineering tools like GitLab and Azure DevOps) via the Admin console; Glean mirrors source ACLs automatically so every user sees only content they can access in the origin system. To migrate from Dust to Glean: for each active Dust connection, locate the equivalent Glean connector, re-authenticate with the same service account, and verify that Glean's ACL sync covers the same permission scopes Dust admins had manually set. To migrate from Glean to Dust: note that Dust has no automatic ACL mirroring — admins must manually configure which folders/channels each Dust Space can access and assign those Spaces to agents.
- Warning: Glean mirrors source ACLs automatically at query time; Dust requires admins to manually scope which folders/channels/pages each Space includes — under-scoped Dust connections silently omit content agents could answer from.
- Warning: Dust's Snowflake and BigQuery connectors allow structured table queries by agents; Glean's engineering connectors (Databricks) cover overlapping territory but direct SQL table-query as a connector feature differs in implementation.
- Warning: Dust beta connectors (Jira, Salesforce, HubSpot, Guru, Linear, Dropbox, Front) may lag behind Glean's generally-available equivalents in sync reliability and feature completeness.