About Dust and why teams compare it
Dust describes itself as multiplayer AI for human-agent collaboration: a place where company knowledge, tools and teams are connected so that agents can be created, shared and run across real workflows. Its home page says it has 70+ connectors and 20+ models, and its pricing page lists custom agents, multi-agent workflows on schedules and triggers, and connecting any tool via MCP. Plans are per seat: a free tier, Pro and Max seats with monthly credit allowances, and a custom Enterprise plan with pooled credits, SCIM, audit logs and single-tenant deployment. Teams compare it with Swfte when they want evaluation and versioned releases around their agents, or want usage counted differently from seats and credits. This page compares the two on published facts and says who should pick which.
Dust sits in the Platform for company agents over your data and tools category. In short: A hosted workspace where teams build and share AI agents connected to company knowledge and tools, priced per seat with a monthly credit allowance. Pricing as published is Listed on the pricing page in euros, excluding VAT: Free €0 with 500 credits for life; Pro €24 per seat per month billed yearly (€30 billed monthly) with 8,000 credits per seat per month; Max €120 per seat per month billed yearly (€150 billed monthly) with 40,000 credits per seat per month; Enterprise custom with pooled credits and volume pricing. Credits do not roll over. It is best for Teams that want to build and share agents over their company knowledge in a hosted workspace, with a published price to start from.
Swfte vs Dust at a glance
| Capability | Swfte | Dust | Checked at |
|---|---|---|---|
| Category | Visual builder for agents and workflows (Studio), on the same platform as the gateway | Platform for human-agent collaboration: agents connected to company knowledge, tools and teams, per Dust | Dust home page |
| Pricing unit and plans | Free (10 workflows, 5 agents, 2,000 executions per month); Pro $39; Team $99 (3 users included); Enterprise custom | Per seat, in euros excluding VAT: Free €0 (500 credits for life); Pro €24 per month billed yearly or €30 billed monthly; Max €120 yearly or €150 monthly; Enterprise custom | Pricing |
| How usage is counted | Executions per month by plan; see the pricing page for model usage terms | Credits per seat per month: 8,000 on Pro and 40,000 on Max. Dust says credit use depends on the model, the complexity of the task and the tools an agent uses, and that unused credits do not roll over. Enterprise has pooled credits | Pricing |
| Plan limits | Free: 10 workflows, 5 agents, 2,000 executions per month | The docs describe the free option as up to 5 users, 3 connectors and 5 Spaces, and Business as supporting up to 100 seats. Enterprise lists unlimited users, connectors and Spaces | Subscriptions docs |
| Integrations | 100+ tool integrations, plus an API for custom systems | The home page says 70+ connectors. The docs name nine native data-source connections (Confluence, Google Drive, GitHub, Notion, Slack, Snowflake, Zendesk, Intercom, Microsoft SharePoint), plus Zapier and API file upload | Data sources docs |
| AI agents and models | Any model reachable through the Connect gateway; agents and multi-agent orchestration in Studio | The pricing page lists 20+ models (GPT, Claude, Gemini, Mistral, DeepSeek), custom agents and multi-agent workflows on schedules and triggers. The docs describe auto models and let admins cap which models and reasoning efforts members can select | Available models |
| MCP | See the MCP gateway page for how Swfte approaches MCP | Admins add a remote MCP server by its public URL, authenticating with automatic OAuth, a bearer token or static OAuth. The pricing page lists unlimited MCP servers on Enterprise. A client-side MCP server is documented as a preview | Adding an MCP server |
| Human approval | Role-based access and human-approval steps in workflows; configurable data-retention policies are designed for, not shipped | The API has an endpoint to approve or reject an action in a conversation message, and agents that work by email send Accept and Reject links when an action needs confirmation. Admins assign user, builder or admin roles | Validate an action (API) |
| Evaluation and versioning | Testing and evaluation: A/B experiments between chatflow versions and safety policies on agents are built; golden sets and regression runs are in development. Versions with promote and rollback for chatflows and workflows; release channels and gradual rollouts are in development | Not verified from the pages we reviewed | |
| Observability and audit | Traces, logs and token-cost telemetry per step | Audit logs are available only on Enterprise, record who did what to which resource, can tell human from agent actions, export as CSV and can stream to Datadog, Splunk, AWS S3, GCP GCS or an HTTPS endpoint | Audit logs docs |
| Hosting | Managed service; customer data at rest in AWS Ireland. Dedicated or private deployment is designed for and scoped as an engagement, not self-serve | US or EU data residency on Business and Enterprise. Single-tenant deployment is listed on Enterprise. A self-hosted option is not stated on the pages we reviewed | Pricing |
| SSO and admin controls | SAML SSO and SCIM are in development | SSO with Okta, Entra ID and Jumpcloud is listed on Business and Enterprise; SCIM provisioning is listed on Enterprise only | Pricing |
| Data use and training | Not listed on this site as a Studio statement; see the trust centre | Dust says: your data is never used to train models, and no data is stored by third-party model providers. Data is encrypted with AES-256 at rest and TLS in transit | Dust security |
| Compliance | No SOC 2 report or ISO 27001 certificate is held and no HIPAA BAA is signed today; a SOC 2 Type I audit is in preparation (see the trust page). The exact posture depends on your use case, jurisdiction and deployment | Dust’s security page says it is GDPR compliant and SOC 2 Type II certified, and that it enables HIPAA compliance. Its trust centre did not render for our fetch, so scope and dates are not confirmed here | Dust security |
| Export and portability | Not verified for Studio on this page | A Conversation API and a JavaScript SDK are listed in the docs index. An export of agent definitions is not verified from the pages we reviewed | Dust docs index |
Dust values were checked against the vendor's own pages on October 6, 2026. Plans, prices and limits change; confirm on the vendor's page before you decide. Where a cell is blank or says "not verified", we could not confirm it from a primary source and chose not to guess.
Who should pick which
Pick Dust if
- You want a hosted workspace where many people create and share agents over company knowledge, with seat-based pricing you can read before talking to anyone.
- You want US or EU data residency on a self-serve plan, which Dust lists on Business, and SSO with Okta, Entra ID or Jumpcloud.
- The systems you need are in its connector list, and you value that it names models from OpenAI, Anthropic, Google, Mistral and others in one picker.
- You want single-tenant deployment, SCIM and audit logs streamed to your SIEM, which Dust lists on Enterprise.
Pick Swfte if
- You want testing, versioned releases and rollbacks around agents rather than editing live agents in place.
- You want the builder and the model gateway on one platform, so routing, failover and cost tracking are shared.
- You prefer usage counted in workflow executions to per-seat credit allowances that reset each month.
- You want human-approval steps in workflows, with Nexus available for agent governance.
- You are comfortable with Swfte’s current gaps: SAML SSO and SCIM are in development, dedicated or private deployment is scoped as an engagement and is not self-serve, and the integration catalogue and connector list should be checked against the systems you need.
What Dust does well
- Pricing is published in euros, with a free tier and Pro and Max seats, so a small team can start without a sales call
- Dust lists 20+ models from OpenAI, Anthropic, Google, Mistral and others, and admins can cap which models members can pick
- Remote MCP servers can be added by URL, with automatic OAuth, bearer token or static OAuth for authentication
- SSO with Okta, Entra ID and Jumpcloud is listed on the Business plan, with US and EU data residency
- Its security page says customer data is never used to train models, and that no data is stored by third-party model providers
Trade-offs to weigh
- Seats are the billing unit and each seat has a monthly credit allowance that does not roll over, so usage-heavy teams need to watch credits as well as seats
- The Business plan is described as for teams of up to 100 people; larger organisations are on Enterprise at a custom price
- SCIM, audit logs, unlimited connectors and MCP servers, and single-tenant deployment are listed on Enterprise only
- Connector counts differ between Dust’s own pages (70+ on the home page, nine natively listed in the docs), so check the exact systems you need
When Swfte is the better choice
When you want a visual builder with testing, versioned releases and approvals on the same platform as a model gateway, and when you want usage counted in executions rather than per-seat credits. Dust is the better fit if you want its seat-based workspace and its published connector set today.
Swfte Studio is for teams that want to build agents and workflows visually and run them as a managed service. It uses the Connect gateway for model access, so routing, failover and cost tracking are shared, and it lists testing and evaluation, semantic versions with rollbacks, and human-approval steps in workflows. Nexus adds agent governance.
Swfte is not a copy of Dust’s shared workspace. Dust publishes its connector set, plan limits and a path to single-tenant deployment. Swfte is earlier on enterprise identity: SAML SSO and SCIM are in development, and dedicated or private deployment is an engagement, not self-serve.
Where Swfte fits
How to migrate from Dust to Swfte
- List your Dust agents
For each agent, note its instructions, the data sources and tools it uses, any MCP servers, its model choice and who uses it. Dust documents a Conversation API and a JavaScript SDK, which can help you record how agents are called.
- Rank by use and by risk
Start with an agent that has clear inputs and a checkable output. Leave agents that depend on many connected sources until you have confirmed that Studio can reach the same systems.
- Record what you have
Copy instructions and tool lists into a document. We do not list an import for Dust agents, and we have not verified an export of agent definitions from Dust, so treat it as a rebuild.
- Rebuild one workflow in Studio
Recreate the trigger, the agent, the tools and any approval step. Studio’s free plan includes 10 workflows, 5 agents and 2,000 executions per month, enough to test one case. Studio lists 100+ tool integrations plus an API for the rest.
- Test on real inputs, then run both
Run 20 to 50 real inputs through Studio’s testing and compare with the Dust agent. Then run both for a while and compare outputs, failures and cost.
- Cut over what fits, and review seats at renewal
Move the workflows that benefit from versioned releases and keep the rest in Dust. At renewal, decide how many Dust seats you still need.
How Dust compares to other alternatives
Other products worth comparing are Glean, Copilot Studio and Lindy. Each has its own page in the alternatives index.
For a full cross-comparison see the alternatives index and the head-to-head comparisons grouped by category.
Sources
Dust facts on this page come from the vendor's own documentation and pricing pages. Swfte facts come from our pricing and product pages. See something out of date? Tell us at sales@swfte.com and we will correct it.
Frequently asked questions about Dust alternatives
Is Swfte a Dust alternative?
For building and running agents with a visual builder, testing and approvals, yes. Dust is a hosted workspace where teams share agents over connected company knowledge, and its connector list and seat-based model suit that way of working. Studio lists testing and versioned releases, which we could not verify on Dust’s pages; check the systems you need against our integration list first.
How much does Dust cost?
Dust lists prices in euros, excluding VAT: Pro is €24 per seat per month billed yearly (€30 billed monthly) with 8,000 credits per seat per month, Max is €120 yearly (€150 monthly) with 40,000 credits, and Enterprise is custom. Swfte Studio lists Free, Pro $39, Team $99 (3 users included) and Enterprise custom. The units differ, so compare on your own workload.
Does Dust have a free plan?
Yes. The pricing page lists a Free tier at €0 with 500 credits for life, and the docs describe it as up to 5 users, 3 connectors and 5 Spaces. Swfte Studio’s free plan includes 10 workflows, 5 agents and 2,000 executions per month.
Can I self-host Dust?
We did not find a self-hosted option on the pages we reviewed. Dust lists single-tenant deployment on Enterprise and US or EU data residency on Business and Enterprise. Swfte Studio is a managed service with customer data at rest in AWS Ireland; dedicated or private deployment is scoped as an engagement, not self-serve.
Can I import my Dust agents into Swfte?
No. We do not list an import for Dust agents, and we have not verified an export of agent definitions from Dust. Record your instructions, tools and data sources and rebuild the workflow in Studio.
Does Dust support MCP and human approval?
Yes to both, with limits. Admins can add a remote MCP server by URL, and Enterprise lists unlimited MCP servers. For approvals, Dust’s API has an endpoint to approve or reject an agent action, and email agents send Accept and Reject links. For Swfte’s approach to MCP see our MCP gateway guide.
Is Dust compliant, and how does that compare with Swfte?
Dust’s security page says it is GDPR compliant and SOC 2 Type II certified and enables HIPAA compliance; its trust centre did not render for us, so confirm scope with Dust. Swfte does not hold a SOC 2 report or an ISO 27001 certificate and does not sign HIPAA BAAs today; a SOC 2 Type I audit is in preparation, and the trust page has the current status. Ask each vendor for its documents and check them against your use case.
Which models does Dust work with?
Dust lists 20+ models, naming GPT, Claude, Gemini, Mistral and DeepSeek, with auto models and admin caps on which models members can select. Studio uses any model reachable through the Connect gateway.
Rebuild one agent in Studio
Pick a Dust agent with clear inputs, rebuild it on the free plan, and compare results on your own data.
Free plan · Managed service