Skip to content
TulipFarm

When Dust fits better, and when TulipFarm does

Dust suits teams whose agents mostly answer questions. TulipFarm suits teams whose agents keep records current, with follow-ups that wait weeks for a reply.

TulipFarm

$50 free

In credits at signup

Dust

24€ a month

Per Pro seat, billed yearly

dust.tt/home/pricing

Where TulipFarm differs

Three differences that show up in one invoice follow-up job, each with a note on how Dust handles it.

  • Agents write to tables, not just query them

    TulipFarm

    Agents update the rows your team uses. A required field stops a row saving half done.

    See Tables

    How Dust handles it

    Table queries let agents read structured data with SQL. Its docs don’t describe editing rows or field rules.

    docs.dust.tt/docs/user-documentation/agents/knowledge/table-queries
  • Chosen actions wait on every run

    TulipFarm

    On actions you pick, the agent shows the exact change and waits for an OK on each run.

    See Agents

    How Dust handles it

    Tools have stake levels. A medium-stake approval, like a Gmail draft, is saved per recipient for all agents.

    docs.dust.tt/docs/user-documentation/agents/tools/gmail
  • Follow-ups that wait weeks and survive restarts

    TulipFarm

    A visual workflow waits up to 30 days for a reply, and a restart won’t redo finished steps.

    See Workflows

    How Dust handles it

    An agent can set a wake-up to check back later. A scheduled wake-up runs at most 32 times.

    docs.dust.tt/docs/user-documentation/agents/tools/wake-ups

Feature comparison

Dust’s column comes from its own site and docs. “Not published” means Dust doesn’t say, not that the feature is missing.

Feature comparison
FeatureTulipFarmDust
Workspace and data
Built-in tables for business recordsYesField rules reject bad values before they savePartialAgents query tables with SQL; no field rules described
Import CSV or XLSX into a tableYesPartialCSV uploads become tables agents can query
Files the team sharesYesShared files and projectsYesPods hold shared conversations, tasks and files
Company knowledge that agents searchYesSearches projects and filesYesConnected sources and folders, permissioned by Space
Agents
Saved agents with reusable skillsYesYes
Agents on a scheduleYesUp to once an hourYesFrequency described in plain language
Run codePartialJavaScript or Python in a workflow stepYesComputer runs code in a temporary workspace
Agents use a web browserNoSearches and reads the web; no browserPartialSearches the web and reads up to 10 results
Automation and approvals
Workflow editorYesOne trigger, then ordered stepsNot published
Webhook triggersYesYesGitHub, Jira, Linear, Zendesk, Fathom or custom, with a daily run limit
Approval before an action runsYesShows the exact changeYesSet by each tool’s stake level
Runs resume after a restart without repeating finished stepsYesNot published
Integrations
Built-in app connectionsPartialSlack, GitHub, Google Workspace, Jira, Confluence and TrelloYesConnectors to 20+ data sources; Business plan connects up to 3
Custom MCP serversPartialA manager reviews and enables each toolYes5 remote servers on Business
Agents in SlackYesMention an agent in a channel; access is granted per channelYesMention @Dust in a channel or thread, or message it directly
Security and deployment
SSO and SCIMNoSign in with Google or an email linkYesSSO on Business; SCIM on Enterprise
Audit logsPartialRecords who changed memberships; no audit exportYesEnterprise, with CSV export
SOC 2 Type IINoYes
Personal chats hidden from adminsYesNot published
Run in your own cloudPartialOpen source (AGPL v3.0); self-hosting is unsupportedPartialSingle-tenant deployment and US or EU residency on Enterprise

When to stay with Dust

Dust is the better choice if any of these sound like your team.

Your agents mostly answer from company knowledge
Dust connects Slack, Notion, Google Drive, GitHub and more, with access set by Space.
You want to choose the model for each agent
Dust offers 20+ models from OpenAI, Anthropic, Google, Mistral and others.
Your agents need to crunch files
Dust’s Computer runs code to clean spreadsheets and build decks and reports.
You need SOC 2, SSO or EU residency
Dust has SOC 2 Type II and SSO on Business, with EU residency and single-tenant deployment on Enterprise.

Pricing

Dust prices are in euros and exclude VAT. Free seats get 500 credits for life. Checked on 20 October 2026.

TulipFarm

$50 free

In credits on your first signup. Paid plans aren’t public yet; see the pricing page.

Dust

24€ a month

Per Pro seat billed yearly, or 30€ monthly, with 8,000 credits a seat. Credits don’t roll over.

dust.tt/home/pricing

Switching from Dust

TulipFarm can’t import Dust agents or conversations. Move one job at a time and keep Dust running until you’re happy with the result.

  1. Pick one job that changes records

    Choose an agent whose answers end up typed into a sheet, such as an invoice chase.

  2. Bring the sheet in

    Export the sheet or CSV it queries and import it into a TulipFarm table with field rules.

  3. Copy its instructions into a skill

    Paste the agent’s instructions into a TulipFarm skill and choose which actions wait for approval.

  4. Run both, then switch

    Run both for one cycle and compare the results. Your first signup includes $50 in free credits.

Questions about TulipFarm and Dust

Is TulipFarm a Dust alternative?

Yes, for teams whose agents need to change records, not only answer questions. TulipFarm stores business records in editable tables with field rules, and for the actions you choose, an agent shows the exact change before a person approves it. Dust focuses on agents grounded in company knowledge.

What is the main difference between TulipFarm and Dust?

Dust connects agents to knowledge in your other tools, organized into Spaces, and lets them query tables with SQL. TulipFarm keeps the records themselves in tables that agents and workflows update, with approval on the actions you pick.

Can TulipFarm do everything Dust can?

No. Dust has more connectors, a choice of 20+ models, a Computer tool that runs code on files, SSO with Okta, Entra ID or JumpCloud, SOC 2 Type II, and single-tenant deployment on Enterprise. TulipFarm has editable tables with field rules, a visual workflow builder, and runs that resume after a restart.

How do approvals in TulipFarm differ from Dust?

Dust sets approval by each tool’s stake level. For a medium-stake action like a Gmail draft, one approval is saved per recipient and applies to every agent. In TulipFarm, each chosen action waits for approval on every run, and the person sees the exact change.

When should I just use Dust?

Choose Dust if your agents mostly answer questions from company knowledge, you want to pick models per agent, or you need SOC 2, SSO or EU data residency today.

Can I move my Dust agents to TulipFarm?

Not automatically. Rebuild one agent at a time: export the sheet or CSV it queries and import it into a table, paste its instructions into a TulipFarm skill, and run both until you trust the result.

How much does TulipFarm cost compared with Dust?

Your first TulipFarm Cloud signup includes $50 in free credits. Paid plans aren’t public yet; see the pricing page to ask for more credits or apply for student, NGO or startup support. Dust Pro costs 24€ a seat a month billed yearly, or 30€ monthly, with 8,000 credits a seat that don’t roll over. Max costs 120€ a seat billed yearly, and prices exclude VAT.

Does TulipFarm work with Slack and Google Workspace like Dust?

Yes. In Slack, people mention an agent in a channel you grant, and it replies there. Each person can also connect their own Google account for Gmail, Calendar, Drive, Docs, Sheets and Slides. Reading needs no approval; any change in Google waits until a person approves the exact request.

Try one Dust job in TulipFarm.

Sign up for TulipFarm Cloud and rebuild the agent whose answers end up in a spreadsheet. The research preview is open to everyone, and your first signup includes $50 in free credits.

Not sure TulipFarm connects to your tools? Contact us and tell us what your team uses.

Sources

Last reviewed 20 October 2026.