The desktop

If your device has a screen, the Kiki desktop is where you and the agent meet. It looks calm and minimal because it isn't a pile of windows — it's a view of the work happening on your computer.

The prompt bar

The main way you start anything. It's not a search box — type or speak a goal and Kiki begins a session:

Tidy my downloads folder and group screenshots by month.

You're describing an outcome ("I want this done"), not launching an app ("open the file manager").

The status bar

A live view of every session at once — what each one is doing and how far along it is:

[Editing video] [Researching… ] [Export 73%]        🎤  ⚡84%  14:32

Tap any of them to jump to that session. Kiki updates this itself, so you always know what's in flight.

The launcher

The system menu. Its entries are sessions and activities, not a grid of apps. From it you can start something new, jump to a running or parked session, see what tools are available right now, and reach system settings.

You and the agent share the same controls

Everything you can do from the desktop, the agent can do too — it can type in the prompt bar, update the status bar, and open the launcher. There's no separate "AI mode"; you're both driving the same machine. That's why you can hand a task off and take it back seamlessly.

Talk to it, too

You never have to touch the desktop to use Kiki — voice works everywhere. See Talking to Kiki.

No screen? No problem

Kiki runs perfectly headless on servers and edge devices. There's no desktop there — you drive it from the dashboard or mobile app, or over SSH.

Next: Who's in control.