Autopilot & approvals
How Yuki asks before it acts, shows its work, and lets you set how far it can go.
An assistant that can act is only useful if you can see what it did and stop it doing more. This section is how that works. Agent actions are rolling out — the controls below are the shape they take.
It asks first
Anything consequential — sending an email, making a booking, spending money — produces an approval card before it happens. The card is built from the action itself, not from what the assistant says about it, so what you approve is what runs. Approve and decline are the same size.
It shows its work
The Activity tab lists everything Yuki has done, by day, with what it touched. Tap any entry to open the real record. Where an action can be reversed, undo is there; where it genuinely cannot — an email that has already gone — Yuki says so rather than offering an undo that does nothing.
Then you choose the leash
| Setting | What it means |
|---|---|
| Ask for approval | Recommended. Nothing consequential happens without your tap. |
| Auto-approve this task | Skip approval for the one thing you are doing right now. |
| Auto-approve below a limit | e.g. up to €100 for rides, €50 for groceries. |
| Auto-approve for this trip | e.g. hotels and transport, while you are away. |
Connections
Gmail, Google Calendar, Contacts and your device calendar connect independently, and each is separately set to read-only or allowed to act. Read-only is genuinely read-only: Yuki can answer from it and cannot do anything with it.