Every year Android evolves: background behavior, permission rules, and Play Console requirements. The safest approach is incremental migration: upgrade the toolchain, run regression, then expand scope.

Suggested path

• Upgrade Kotlin/AGP first; fix build and lint issues.
• Run tests and smoke-test critical flows.
• Only then bump targetSdk and handle behavior changes.

Users are increasingly sensitive to permissions and tracking. If your app asks for more access than it needs, installs and retention suffer. Request permissions at the right moment and explain why.

• Ask only when the user triggers a feature that requires the permission.
• Provide a fallback UX when users deny access (don’t block the whole app).
• Data retention: keep less, delete on request.

On-device AI is most valuable when it improves one step in a workflow: suggestions, summarization, search, or classification. Start small, measure impact, then expand.

Implementation principles

• Prefer smaller models with low latency and reasonable battery impact.
• Always provide a manual path when the AI is wrong.
• Don’t send sensitive data off-device unless it’s truly necessary.

Many apps still look like “stretched phone UIs” on tablets and foldables. A few layout rules and sensible breakpoints can dramatically improve UX: list-detail layouts, navigation rails, and better spacing.

• Choose layouts by window size classes, not device models.
• Design a list-detail flow for wide screens.
• Test rotation, split-screen, and multi-window.