> Some Trylle features use artificial intelligence and machine learning, and several of them rely on third-party AI providers (acting as our sub-processors) to operate. When you or a member of your organization invokes one of these features, the relevant context — which may include your source code, diffs, file contents, issues, pull requests, commit messages, and related repository content and metadata — may be sent to and processed by these third-party providers in order to generate a response. This processing happens at your direction: it occurs only when you or your organization triggers the feature, and it is scoped to the repositories and resources that feature is authorized to access.

> You can control whether these features are enabled through your organization and repository settings (for example, Settings → AI), and disabling a feature stops the associated processing going forward.

This playbook feels so familiar.