About
A workbench for the model, with AI in it.
Cairn is a model-based systems engineering workbench built around AI from the start. You describe a physical system; Cairn interviews you, then builds a structured model with you: its parts, requirements, interfaces, behavior and verification, each traced to the next. You read it through twelve lenses and change it by hand or by asking. The AI only ever proposes, and nothing changes until you say so.
Where it stands
In early access, before launch.
Cairn is at version 1.20.0-beta. It is free during early access: every account gets 50 AI actions every 30 days, or unlimited use with your own Anthropic key.
Version 2.0 will be the public launch, the first version people can depend on. Until then the way a model is stored may still change; projects you already have are brought up to date when they open.
How it works
Three things it won't bend on.
The model is the work.
Not a transcript. Every part, requirement and interface is a record you can open, edit, trace and export, and the model's own checks keep it whole.
Every change is reviewed.
The AI proposes changes; you see each one in Change Review and apply what you tick. History keeps them all, so any change can be undone and the model restored to any point.
Your model is yours.
Projects live in your browser. Backing up to your account is your choice for each project, and export takes the whole project with you. The AI is Claude: Claude Haiku 5.5 routes each request and Claude Sonnet 5.5 writes the proposal.
Next
The road to 2.0.
Cairn is built one area at a time, each rebuilt from first principles and released as its own beta. The release notes are in the version history.
Before launch: the rest of this site rebuilt on the real product, an interactive demo made from the app itself, and plans for after early access.
Contact
Tell us what you're building.
Questions, bugs and ideas go to greg@getcairn.dev. Signed in, Help & Feedback in your profile sends a note straight to us. Or start a model and see for yourself.