Trust
Your learning data is not the product.
Cairn is local-first. Your learner data stays on your machine. AI is optional. Nothing is uploaded to a Cairn account because there is no Cairn account.
What stays local
- Your learner profile and mastery estimates.
- Answers, confidence, hint use, and review history.
- Imported notes, textbooks, and packs.
- The local library and pack exports.
What leaves the machine
Only what you explicitly choose. There are three cases.
The app shows the AI mode. It does not silently switch to cloud AI. Pack downloads use the catalog host, but learner data is not sent with the request.
No account, no telemetry by default
Cairn does not require an account. There is no default telemetry. If optional diagnostics are added later, they will be opt-in, visible, and documented here before release.
Export and delete
You can export your learner data and delete progress from the app. The export includes your pack, learner state, events, decisions, model parameters, and policy. It is your file.
Packs
A pack is content, not learner data. The planned community library is a static catalog. Adding a pack copies it into your local library. The store does not receive your progress.
AI is not the policy
AI can generate hints, feedback, and draft practice. It does not choose the learning path. The adaptive policy is deterministic and explainable.
Open source
Cairn is MIT licensed. The code, decision logic, and data format are public. You can audit them, run them offline, and fork them.
Security
Report security issues privately to the maintainer rather than opening a public issue with learner data.
This page describes the current implementation. It will change as the app changes. The source and release notes are the authoritative record.