Sim vs Dust

Sim is worth evaluating when Dust's knowledge agents need to run inside multi-step automations. Dust builds agents from instructions, tools, and connected data sources, and offers no loop container or branch-and-join execution model. Sim adds a visual workflow canvas with loops, parallel branches, error-handling paths, and human approval steps, started by webhooks, schedules, API calls, or events from connected apps.

It depends on where the agent needs to live. Dust connects agents to sources such as Slack, Notion, Google Drive, Confluence, and GitHub and serves them in its web chat, Slack, and Microsoft Teams, with access scoped by Spaces. Sim's knowledge base combines vector and full-text search over 15 file types with connectors that keep sources such as Google Drive, Confluence, and Slack in sync, and a knowledge agent can be published as a chat interface, a REST API, or an MCP server. Dust suits a managed assistant used mainly in chat; Sim suits a knowledge agent that also has to feed workflows or other systems.

Both are built for teams, but they share different things. Dust shares agents and conversations: people and agents work in a common workspace, access is set by roles plus Spaces, and Pro seats cost $30 or €30 per month ($24 or €24 per month billed annually) with 8,000 credits each. Sim shares the build: several people can edit the same workflow canvas live, workspaces carry admin, write, and read permissions, and the Pro plan is $25 per user per month.

Sim is an open-source AI workspace where teams build, deploy, and manage AI agents visually, conversationally, or with code. Dust is an enterprise AI agent platform where teams build no-code agents connected to company data and tools in a shared, multiplayer workspace, then deploy them to chat, Slack, and other surfaces. Teams considering a switch typically weigh licensing (Sim's core is Apache 2.0 and self-hostable), pricing model, and how AI-native the platform's agent-building experience is.

Sim is built specifically as an AI agent workspace, with native multi-LLM support, an in-editor AI Copilot, and a knowledge base with hybrid vector + keyword search. Dust's standout capability is zero visual/flow layer by design, building only through forms, text, and conversation: Dust's Agent Builder is entirely form and text based, name, description, instructions, model, tools, knowledge, guided by a conversational 'Sidekick' assistant, with no visual canvas at all (its earlier block-based 'Dust Apps' product is deprecated). Agents deploy natively into a shared, multiplayer workspace and out to Slack, Teams, and other chat surfaces. A team that wants agents assembled purely from plain-language instructions and templates, with no drag-and-drop layer to learn or maintain, gets that directly. Teams that do want infrastructure-as-code can also define Skills and agent configurations as files in a Git repository and sync them via an official GitHub Action, with the same PR review and rollback workflow as application code. One documented limitation of Dust is no visual, node-based canvas: agents are configured through forms and text: Dust's agent builder is a form-based, instruction-driven interface (name, description, instructions, model, tools, knowledge), not a drag-and-drop node/flow canvas. Its earlier block-based visual orchestration product, 'Dust Apps', is deprecated: only apps created before October 2025 remain accessible, and creating new ones is disabled.

Sim can be self-hosted via Docker or Kubernetes with an Apache 2.0-licensed core, in addition to a managed cloud-hosted plan. Enterprise features such as SSO and audit logs require an Enterprise license. Dust's self-hosting position: No: the core repository is MIT-licensed and public on GitHub, but self-hosting isn't an officially supported deployment path; Dust is sold and operated only as hosted SaaS.

Sim uses credit-based usage billing (Stripe); bring-your-own-key calls skip Sim's model and tool usage charge, while the base per-run charge still applies. Dust uses per-seat subscription (Free, Pro, and Max seat types on the Business plan, for teams up to 100 people) with a monthly AI-usage credit allocation per paid seat; Enterprise is sales-led.

Security is a like-for-like comparison, not a one-line verdict. Sim: SOC2. Dust: GDPR compliant, HIPAA-capable, SOC 2 Type II; no ISO 27001, PCI, or FedRAMP. Check the Security & compliance rows above for the full breakdown, including SSO, audit logging, and data residency.

Sim: 28 provider integrations (OpenAI, Anthropic, Google/Gemini, Azure OpenAI, Azure Anthropic, Vertex, Bedrock, xAI, Mistral, DeepSeek, Groq, Cerebras, Fireworks, Together, Ollama, OpenRouter, and more), with OpenRouter, LiteLLM, vLLM, and Ollama resolving models dynamically at runtime rather than from a fixed list, so effective model reach extends well beyond the 28 named providers. Dust: agents can be configured with a choice of model (e.g. GPT-4 Turbo, Claude 3, Gemini Pro, Mistral Large) and a Reasoning Effort setting, selectable per agent. Sim also ships native human-in-the-loop approvals, a hybrid vector-plus-keyword knowledge base, and an in-editor AI Copilot that can read execution logs and directly edit the workflow to fix a failed run.

Documented limitations of Dust include no visual, node-based canvas: agents are configured through forms and text: Dust's agent builder is a form-based, instruction-driven interface (name, description, instructions, model, tools, knowledge), not a drag-and-drop node/flow canvas. Its earlier block-based visual orchestration product, 'Dust Apps', is deprecated: only apps created before October 2025 remain accessible, and creating new ones is disabled.; no dedicated pre-deployment evaluation/dataset-testing framework, a gap shared with most agent builders: Dust explicitly says it is 'not a pre-deployment evaluation platform': dataset-based regression testing belongs in CI/CD pipelines and specialized testing tools, and Dust builds observability signals into the agent-builder workflow instead of a formal eval-suite feature. This is a gap most agent builders share, including Sim, whose own Evaluator and Guardrails blocks are per-call scoring/validation primitives rather than a batch golden-dataset eval-suite runner.; self-hosting is not a supported deployment path despite an MIT-licensed core: The core dust-tt/dust repository is MIT-licensed on GitHub, but Dust is sold and operated only as hosted SaaS. There is no documented, supported way to self-host a production Dust workspace on customer infrastructure.; no native spreadsheet-like data table; structured data is queried, not edited: Dust's Query Tables tool lets an agent generate and run SQL over structured sources (CSVs, Notion databases, Google Sheets, Snowflake, BigQuery), but Dust has no native, editable spreadsheet-grid feature with arrow-key navigation and copy-paste, unlike a dedicated data-table product.

There is no automated one-click migration tool between Dust and Sim. Workflows and automations need to be rebuilt in Sim's visual builder, natural-language Chat surface, or API. Most teams start by recreating their highest-value automation first to validate the switch before migrating the rest.