Everything Cerevisor does, feature by feature
Build a team of AI agents on a canvas or in chat, give each one its own model, and keep every run on your machine. Works with the tools you already pay for: Claude Code, Codex, Cursor, Gemini, API models and local Ollama.
What you get
- Canvas and Agents: Drag agents onto a canvas, wire their inputs and outputs, and run the whole graph. Free covers up to 2 workflows and 2 saved worlds.
- MCP for tools and context (Pro): Connect any Model Context Protocol server. Filesystem, Git, your private databases, internal APIs.
- Loop and Enhance (Pro): Iterate a node against its own output until a quality bar is met.
- Scheduler (Pro): Run a workflow on a cron, on a file watcher, or on demand.
- Chat Builder: Describe the workflow you want and the Chat Builder assembles the canvas for you.
- Workshop and Skills (Pro): Author and edit reusable skills. Free runs skills already on disk.
- Web search and cloud Cursor (Pro): Built-in web search tool and cloud-hosted Cursor agent runs. Local Cursor runs work on every plan.
- Per-agent providers: Mix providers in one workflow: Claude Code for one agent, Gemini for the next, a local Ollama model for a third. Included on every plan.
- A check before every run: See every agent's role, model and permissions before a run starts. Included on every plan.
- Roles for everyday work: Presentation builder, Excel expert, meeting-notes summariser, research-brief writer and more, plus Outlook mail and calendar tools on Windows. Included on every plan.
Built for people who own their stack
Cerevisor runs on your machine and keeps its usage data there, sending none of it to Cerevisor. Bring your own provider keys or plans. Workflows live as plain files you can version, share, and audit.