Windows desktop app

Talk to your computer.
It actually does things.

Overview listens, then acts. Files, applications, windows, and code, driven by voice from one small window that stays out of your way.

Try it now

Talk to the orb

OVERVIEW 09:24:11

Listening

Talk to it, or try dragging a file onto it. Voice needs mic access, which your browser will ask for.

Drop in files and folders
Controls apps and windows
Connects GitHub, Slack, Drive, and more

No commands to memorise

Say it the way you'd say it to a person

"Find every invoice PDF from last month."

"Put this window on the left monitor."

"Write a Blender script for a low-poly tree."

"How much space is left on my drives?"

"Open the project I was in yesterday."

"Screenshot this and copy it."

It reads a command sheet for each application it knows, so it runs real commands instead of guessing where to click.

New: it remembers, and it shows real data

Notes, standing instructions, and shortcuts it actually keeps

"Remember that the Q3 numbers live in the finance folder."

"From now on, always show me storage as a chart."

"Whenever I say movie night, dim the lights and open Netflix."

"Show me the SpaceX stock."

"What's the weather doing this week?"

"What did I ask you to remember about the lease?"

Notes live in real folders it organizes itself, standing instructions shape every reply from then on, and a stock or weather question pulls real numbers from a real source - never a guess - drawn as an actual chart you can click through, not just read back to you.

Watch the whole project, not just one file

A live file tree for whatever it's building

Calculator

Live content


        

Files

Every file it touches shows up here as it's read, written, or edited - the same panel from inside the app, not a mockup of it.

Writes essays and documents too

Not just code - real writing, drafted live

the_case_for_slow_mornings.txt

    

Essays, emails, cover letters, first drafts - dictate the idea and it writes the document, not just a paragraph to copy into one yourself.

Pricing

Bring your own key.

Overview runs on your own Gemini, OpenAI or Claude key, and that provider bills you directly for what you use. Gemini has a free tier, so you can start without paying anyone. Every plan includes AI agents you can deploy to work for you - Free runs one at a time, Basic five and Plus thirty - and Basic and Plus add Connections: GitHub, Google, and your other apps and accounts.

Free

$0

No billing

  • Bring your own Gemini, OpenAI or Claude key
  • Conversation, writing, coding, and everyday actions
  • Billed by that provider directly, not by us
  • 1 AI agent deployed at a time
Download

Cancel any time, and keep access until the period you paid for ends.

Questions

Before you download

What are agents?

Assistants you set up for a job - research, writing, checking something every hour - and deploy to work on their own while you carry on. Each uses your own AI key, and anything that sends, deletes or spends money asks you first unless you tell it not to. Free runs one at a time, Basic five and Plus thirty, and you don't need an account to use one.

Does my voice get uploaded?

No. Speech recognition and the voice it speaks back with both run on your own machine. Only the text of a request reaches a model.

Can it delete something by mistake?

Deleting is the one action that stops and shows you what it's about to do. It also only reaches folders you've allowed, and deletes go to the recycle bin.

What do I need to run it?

Windows 10 or 11, 64-bit, and a microphone. The installer is a few hundred MB - the speech recognition and voice models that run locally are the only thing it downloads, nothing else to pick or configure.

Does it work offline?

Speech recognition and the voice it speaks back with both run on your machine either way. Actually talking to it and having it act needs an internet connection, since each request goes to the provider of your own key.

One window. Ask out loud.

Windows 10 and 11