What Sim is
Sim is an open-source platform for building and deploying AI agent workflows. Over 100,000 builders use it, and the project has 29,000+ GitHub stars.
You build agents on a visual canvas by connecting blocks for models, tools, logic, and data. It works with every major model provider, including OpenAI, Anthropic, and Google, plus local models through Ollama. Over 1,000 integrations connect agents to the tools teams already run.
How it works
Drag blocks onto the canvas and wire them together. An agent block calls a model and can use any tool you give it. Logic blocks handle branching, loops, and conditions. Knowledge bases handle RAG with chunking and vector search built in. Tables give agents memory and state that persists across runs. Multiple people can edit the same workflow at once.
Deploying
A finished workflow ships as a REST API, a scheduled job, a webhook, a chat interface, or an MCP server, which means external AI assistants can call your Sim workflows as tools. Sim also works the other direction, importing any MCP server as a tool inside your agents.
What teams build
Customer support triage, sales research and lead enrichment, internal document Q&A, data pipeline automation, and back-office operations.
Why it's different
Most workflow tools were built for deterministic automation and had AI added later. That gap shows up as soon as you need an agent to reason across steps rather than follow a fixed chain of triggers. Sim was designed for agents from the start, so model calls, tool use, and multi-step reasoning are native.
Sim is Apache 2.0 licensed with a patent grant, not fair-code or source-available. You can self-host the whole platform on your own infrastructure, which matters for teams with data residency requirements. The free tier is fully functional rather than a limited trial.
Enterprise deployments add SSO, SOC 2 compliance, access control, self-hosting, and dedicated support.
