How we use AI at Cyclop

• Auditing code and configurations, to point out bugs, security weaknesses and possible improvements.
• Helping write and refactor code, tests and packaging, under an engineer’s direction.
• Drafting documentation and translations (Cyclop OS ships in 39 languages), which are then checked.
• Analysing logs and reproducing problems faster.

• An engineer reads and approves every change before it is merged.
• Every change is built from source and runs through automated checks: test suites, and side by side comparisons against the previous version when we rewrite a component. Our installer components rewritten in Rust, for example, had to give identical results to the old code on every test case before they shipped.
• It is then installed and tested on real machines and on virtual machines before release.
• Nothing ships automatically. Updates are grouped into regular patches, every week and two weeks at most, and a package under test never reaches the public repositories before patch day.

The engineer who approves a change is responsible for it.

• AI does not run on your computer as part of Cyclop OS, and Cyclop OS does not send your files or your activity to an AI service.
• AI tools never get access to customer machines or customer files.
• AI does not decide what ships. No change is published without a person approving it.

• Privacy and telemetryEvery connection the system makes, and what it sends.
• SourcesOur public repositories and source packages.

It lets a small team audit more code, more often, catch problems earlier and ship security fixes faster, while keeping a person accountable for every line that reaches your machine.