Sim is an open-source AI agent builder for teams that design agent workflows on a visual canvas and run them in its cloud or on their own servers.
Sim turns a chain of model calls, tool calls and business rules into a workflow that runs when an event arrives. Its maker is Sim Studio, Inc., which started in 2025 and is based in San Francisco. The core is open source under the Apache 2.0 license. Teams use the hosted version or deploy their own copy with Docker or Kubernetes. Work starts on a canvas where each step is a block, though agents can also be built through chat or in code. The Agent block calls a model from OpenAI, Anthropic, Google, xAI or Groq. It can return output shaped by a JSON Schema and call custom JavaScript tools or tools from any MCP server. Knowledge bases ground answers in uploaded documents, the Router block picks a path, and the Workflow block calls a child workflow. The Guardrails block checks content for valid JSON, PII or ungrounded answers, and the Human in the Loop block pauses a run until someone approves it. Engineering and operations teams use Sim to build agents visually and still ship them like software, with deployment versions, APIs and Logs that show every trace span and its cost. Product teams publish an agent as a REST endpoint, a shareable chat page or an embedded widget. Individuals and small teams start on a no-cost tier, then buy seats on plans that differ in credits, workspaces and run limits. Organisations that need single sign-on, access control, SOC2 compliance or Self Hosting buy the Enterprise plan through sales. An AI Slackbot that answers staff questions from a knowledge base is a common first build. IT and support teams route incoming requests by topic and urgency, with a person approving the replies that matter. Operations teams pull fields from documents arriving in Gmail or Notion into a table, then pass the record on to Salesforce. Each run starts from a Webhook, Schedule or RSS trigger, and a bad release rolls back to an earlier version with Promote to live.
Pricing
View plans-
Free
Free
- Free forever, $0 in both the monthly and annual views
- 1,000 one-time credits with no weekly refresh
- 1 workspace and no teammate invites
- 30 days of log retention, 5 GB file storage
- 5 tables with up to 50,000 rows each
- 10 concurrent executions; 50 sync, 200 async and 30 API endpoint runs a minute
- 5 minute sync timeout and 90 minute async timeout
-
Pro
$25 per month
$25 per user a month billed monthly; $21 per user a month billed annually (15% off)
- 6,000 credits a month plus 2,000 a week
- 3 workspaces with teammate invites
- Unlimited log retention, 50 GB file storage
- 100 tables with up to 100,000 rows each
- 50 concurrent executions; 150 sync, 1,000 async and 100 API endpoint runs a minute
- 50 minute sync timeout and 90 minute async timeout
- No Sim Mailer, KB Live Sync, Sandboxes or Slack Connect; no SSO, access control, SOC2 or Self Hosting
-
Max
$100 per month
$100 per user a month billed monthly; $85 per user a month billed annually (15% off)
- 25,000 credits a month plus 4,000 a week
- Unlimited workspaces
- 500 GB file storage, 1,000 tables with up to 500,000 rows each
- 200 concurrent executions; 300 sync, 2,500 async and 200 API endpoint runs a minute
- Adds Sim Mailer inbox, KB Live Sync, Sandboxes and Slack Connect
- No SSO, access control, SOC2, Self Hosting or dedicated support
-
Enterprise
Custom pricing
- Custom price through sales
- Custom credits, refresh, limits, storage and timeouts
- Access control, SSO, SOC2 compliance and Self Hosting, offered on this plan only
- Dedicated support
Features
- Included: Visual agent workflow canvas
- Included: Custom tool and function definitions
- Included: Model provider choice per agent
- Included: Retrieval-augmented knowledge bases
- Included: Multi-agent handoff and delegation
- Included: Conversation state and checkpointing
- Included: Structured output schema validation
- Included: Input and output guardrails
- Included: Human-in-the-loop approval steps
- Included: Built-in execution tracing
- Not included: Prompt versioning in the builder
- Not included: Agent evaluation and test suites
- Included: MCP server support
- Included: Webhook and event triggers
- Included: Code-first SDK
- Included: Deploy an agent as an API endpoint
- Included: Embeddable agent chat widget
- Included: Self-hosted agent runtime
Additional Features
- Agent block with a model chosen per step
- Custom tools written as JavaScript functions
- Knowledge bases with search by meaning
- Router block for model-chosen paths
- Workflow block for calling child workflows
- Guardrails block for JSON, regex, PII and grounding checks
- Human in the Loop block with approval portal
- JSON Schema response format on agent output
Best for
- Engineering teams putting AI agents into production behind an API
- Operations teams automating IT and support ticket triage
- Teams that want an open-source agent builder they can self-host
- Companies that need a person to approve an agent's action before it runs
- Teams mixing models from OpenAI, Anthropic, Google and others in one workflow
- Product teams adding an agent chat to a website or app
- Enterprises that require SSO and SOC2, which come on the Enterprise plan
- Individuals trying agent workflows on a no-cost tier before buying seats
Use cases
- Building an AI Slackbot that answers from company documents
- Triaging IT requests and routing them to the right queue
- Triaging support tickets by topic and urgency
- Pulling fields out of incoming documents into a table
- Exposing an agent as a REST endpoint for another application
- Embedding an agent chat on a website
- Holding an agent's reply for human approval before it is sent
- Running a workflow on a schedule or from an RSS feed
- Grounding agent answers in an uploaded knowledge base
- Publishing workflows as MCP tools for other agents
- Rolling a workflow back to an earlier deployment version
Screenshots & Videos
Explore Sim in action
User Reviews
Compare with Similar Software
Click to compare side by side with Sim
