Use case · For privacy

AI for files you’d never paste into a chatbot

Tax returns, medical bills, a lease, the kids’ school records: some files shouldn’t end up on an AI company’s servers. Run an open model on your own Mac and give the job to an Agent who thinks with that model. Ada reads the files on your hardware and tells you what needs your attention. If you like, the place your workspace keeps its data can be on your own hardware too.

“The files I read, and the model I read them with, stay on this Mac.”

Ada

Ada worked through 300 pages of bank statements, receipts, and medical bills, and not one page went to an AI company.

Tax weekend on the family Mac Studio
  1. Sat 9:00 am

    Drags a folder of bank statements, receipts, and last year’s return onto Ada’s computer.

  2. Sat 9:24 am

    Reads every page with an open model running on the Mac itself and builds a checklist of what’s missing.

  3. Sat 9:40 am

    Totals the year’s medical bills and flags two that the insurer hasn’t paid yet.

  4. Sat 9:41 am

    “Your home office costs look different from last year. Want me to draft a note for your accountant explaining why?”

How it works

The pieces behind the story

Where it runs

Ada works on a walled-off computer inside your Mac, and the model runs on the same machine, so what Ada reads never goes to an outside AI company.

Who does it

Ada reads and organizes. You decide what leaves the house, such as that note to your accountant.

What you control

You decide who can open Ada’s work, and not even our support team can look without your permission.

Your Agent and the model run on your own equipment, and your workspace’s saved data can stay there too. A model you run yourself has no per-token bill.

Set it up

Getting set up

  1. 1

    Install Geyser Host on your Mac and switch on “Run local models here”.

  2. 2

    Choose a model and give it to an Agent.

  3. 3

    To keep your workspace’s saved data on the Mac as well, talk to our team. During the preview, we set that up with you.

Keep exploring

More use cases

Browse all use cases
Make a start

Start with one Agent and one real job

Set up a workspace, give an Agent something useful to do, and see how far they get.