Skip to content

Cairn Documentation

Quick Start

Get from zero to a working system model in under five minutes.

Prerequisites

A modern browser (Chrome, Firefox, Safari, Edge) and a free account. The free tier includes a limited number of nodes and AI calls — enough to explore the platform and build a small model.

Step 1: Describe Your System

Open the app. With no systems yet, the page asks What are you building?; otherwise click Begin a new project. Type a plain-language description of your system, as you would tell a colleague. It can be rough:

"An autonomous delivery rover with LIDAR navigation, hub motors, and a lockable cargo compartment"
"A portable weather station with solar power, cellular connectivity, and temperature/humidity/pressure sensors"

Click Start the interview (or press ⌘Enter). Back up to my account as I go is ticked, so you can carry on from another device; untick it to keep the project on this computer. Start from a blank model skips the interview.

Step 2: The Interview

Cairn asks two to four questions a round, one at a time, most important first, for up to three rounds. Pick an option (click it, or press its number and Enter), answer in your own words, or choose Not sure. Assume for me: Cairn assumes something reasonable and marks what follows from it as an assumption. Back and Change return to an earlier answer.

The questions explain themselves. Each option has a line on what choosing it means, and the one you choose opens on its consequence for the design and what it adds to the model: an actor, a requirement, or the part of the architecture it shapes. Why this matters says what a question decides, and Not sure says what Cairn will assume before you choose it. Point at an option, or choose it, and Your model so far shows what it would add, drawn in the sketch, before you go on.

While Cairn reads your description, and while it writes each next round, the page shows what it is writing as it writes it: what it understood, the questions, the sharpened description, and the actors and requirements, with your own words or answers beside them.

Beside the questions, Your model so far shows the system's name (rename it there), how clear the idea is, a sketch, and what will go into the model: the system's actors (the people and systems it deals with) and its requirements, each marked You said or Assumed, with a switch to keep it or not. What you said is kept; assumptions stay out unless you keep them.

After each round, Here's what your answers added leads with what your answers did (added, changed, confirmed), shows each actor and requirement the round brought in or changed with the answer it came from, and marks the same ones in your model beside it. Facts Cairn only rephrased are named in one line, as reworded with their meaning unchanged. A fact keeps its identity from round to round, so a renamed one keeps your choice to keep it or not. An assumption says where it came from: your description, an answer, or a question you answered Not sure. Ask the next round's questions, or Build now; you can build at any point.

Save and finish later keeps the interview where it is. It is saved as you go, so it is waiting under Pick up where you left off on this computer, and on any device signed in to your account if it is backed up.

Step 3: The First Build

When you build, the build tells you what it is doing, with your own names:

  • The project gets a system node carrying your description.
  • The actors and requirements you kept are added in one ChangeSet: each actor as an external node, each requirement on the system, with a rationale saying where it came from.
  • When an image provider is set up, a first concept image of the system is drawn, and shown as soon as it is ready.
  • The Architect proposes the system's parts and the interfaces between them and with its actors, in one call.

Nothing in the proposal is added until you say so. Each part and interface has a tick, and the sketch shows what you left out; an interface whose other end you left out can't be added without it. Add applies what is ticked, Review every field in Change Review opens the full review, and Open without them leaves the proposal waiting: the Overview's From your interview card offers it again, with the next steps the interview points to. The interview itself (your answers, each fact and what it became, how the proposal was settled) stays with the project, as the Origin entry in History.

Day One

The first run ends on a page about your system: its concept image (or a drawing of its parts), what you built and from how many answers, and how many requirements you said and how many Cairn assumed. Your cairn draws the model's completeness as stones, one for each of what it is, what it does, how it connects and how you'll know it works; these are the Completeness lens's checks. Your next moves names three, each with why it matters; on a computer, Ask Cairn opens the workspace with the request ready in ⌘K. The Architect's own suggestions follow, marked as the AI's, and See it in Cairn previews your model in Architecture, Requirements, Completeness and Visuals. Open the workspace carries on.

Building On: Your Cairn

The cairn comes with you into the workspace, so every change shows what it added and leads to the next:

  • The status bar keeps the system's cairn and how many checks it meets; click it to open Completeness on the system.
  • Change Review names the checks the operations you ticked would meet, on the system and on any part that was already there, before you apply them.
  • After a change that meets a check that was open, a card in the corner shows the stone that grew, any stone it completed, and the next move, with Ask Cairn. When every check on the system is met, it says so and points to the part with the most left to do. Undo and restore raise no card.
  • The Completeness lens draws the selected node's cairn and starts with its next move; From your interview on the Overview draws the system's, with why each move matters.
  • Your systems shows each system's cairn and its next move, to pick up where you left off.

Starting on a Phone

The interview and the first build work on a phone, saved to your account as you go. Keep the screen open while it builds; if the phone puts the page to sleep, the request goes out again when you come back. Day one opens with Email me a link for my computer, which sends a sign-in link to your own address that opens the new system; the workspace (diagrams, lenses and the AI) needs a bigger screen, and the next moves wait for you there. On a phone, Your systems lists everything in your account, with interviews to carry on first, and an interview paused on either device continues on the other.

Step 4: Explore with Lenses

Your model now has structure. Select any node in the system tree (left sidebar) and explore it through different lenses (top tabs):

  • Overview — Dashboard with health metrics and children
  • Requirements — Functional, performance, and safety requirements
  • Architecture — Child nodes as blocks with their interfaces at every depth, an N² chart, and the interface list
  • Behavior — State machines for operational modes
  • Causality — What prerequisite technologies does this need?
  • Completeness — What's missing from this part of the model?

Each lens shows the same model from a different perspective. The data doesn't change — only how you view it.

Step 5: Refine Your Model

Three paths handle different kinds of change. Pick whichever matches the work in front of you.

Command palette for AI-assisted changes — large-scale decomposition, generating multiple requirements at once, drafting a state machine. Press ⌘K (or Ctrl+K on Windows) and type a natural-language request scoped to your selected node:

"Decompose the power subsystem into battery, charging, and power distribution"
"Add thermal requirements for outdoor operation"
"Generate a state machine for the navigation controller"

The AI produces a ChangeSet. Review it operation-by-operation before applying.

Inspector panel for direct edits — renaming a subsystem, fixing a description, adjusting a requirement's priority. Click the field you want to change and type. Edits commit immediately.

Right-click menu for contextual operations — adding a child, deleting a node, connecting two nodes. Right-click any item in the system tree or the Architecture canvas. The menu lists the operations available in context, and deletions show a cascade preview of what else will be removed.

What's Next

You now have a working model. From here you can:

  • Dive deeper into any subsystem (right-click → Decompose with AI, or select it and press ⌘K)
  • Generate visual renders of components (Visuals lens)
  • Open the Simulation notebook to analyze power, mass, or a mission. Saved notebooks stay with the project; a stale one can be rerun without losing the earlier result
  • Export to PowerPoint or Word for stakeholder review