Maya — 38, a nurse practitioner in Kirkland. Two kids, one aging father, and a calendar she did not choose.
Maya is a composite written to make the product concrete. She is not a real customer and nothing here is a testimonial.
01
A thousand strangers already have a file on her
Maya has never seen any of it. An airline knows her seat. A pharmacy knows her refills. Eleven retailers know her size, and four of them are wrong. She has never been shown a single one of these files, cannot correct the wrong ones, and has never been paid a cent for any of them.
Everybody has a copy of Maya except Maya.
Underneath. This is the status quo the whole company exists to invert: a thousand shadow profiles, each assembled without her, none of them hers.
02
She is given an agent, and it is hers
It takes about ten seconds. She picks a name. Nothing to pay, no card, no trial that expires. The agent installs on the MacBook she already owns and runs there — not in our building, not on our servers.
Given, not sold. It lives on her machine, not ours.
Underneath. Free to every American citizen, permanently. The revenue never comes from her, so it cannot come from her data, her attention or her contacts either. It runs on hardware she owns, using subscriptions she already pays for.
03
It goes and finds out who the world thinks she is
While she makes coffee, her agent reads the public internet about her — every search engine, every answer engine — and comes back with an organised picture. Most of it she recognises. Some of it is out of date. Two things are simply wrong, and one of them has been wrong for six years.
The file everyone else has, finally handed to the person it is about.
Underneath. The intelligence agent runs a research plan across public sources and consolidates the result into a Personal World Model: everything the world has already worked out about her, in one place, for the first time.
04
She corrects it, and it becomes something the world could never know
She fixes the two wrong things in under a minute. Then she adds what no search engine could have found: that she will not fly before 9am, that her father's cardiologist matters more than her own dentist, that she loathes being upsold, that Thursday is the only night she cooks.
A world model is what they figured out. A knowledge model is what she decided to tell.
Underneath. The Personal World Model becomes a Personal Knowledge Model. Nothing is inferred that she did not say; everything derived is shown to her and is hers to change or delete. This is the difference between a profile that guesses and a profile that knows.
05
She turns it on, one field at a time
A hotel wants four things for a stay. Her agent grants three and refuses the fourth. A retailer wants her sizes and offers 15% for the privilege — she takes it. An insurer asks for something she is not willing to give, and simply does not get it. Every one of these leaves a receipt she can read.
The world subscribes to her, instead of copying her.
Underneath. Her Personal Knowledge Model becomes a subscription. Counterparties subscribe to named scopes for a named purpose, per-field and revocable. Sensitive scopes are never bundled. Each is granted on its own or not at all.
06
The most valuable thing she owns turns out to be where she is
Maya lands at 11:40pm. Her hotel already knows, because she let it know for exactly tonight — so the room is ready, the late check-in is handled, and nobody asks her to fill in a form at midnight. At 7am the grant expires by itself.
Stale data is a record. Live data is a service. She owns both.
Underneath. Real-time location is the most undervalued asset a person holds, precisely because it is worthless stale and enormous fresh. Granted narrowly, for a stated purpose, for a stated window, it does more work than any static profile field, and it expires without her having to remember to revoke it.
07
Her agent gets smarter because she chose to make it smarter
Maya's agent runs on her MacBook Pro and handles almost everything there. When something genuinely needs more, a hard piece of research or a long reasoning job, it asks, and she decides whether it goes up the ladder.
Her agent is as powerful as the computer she owns and the keys she hands it. Both are her decision.
Underneath. How capable her agent is depends on two things she controls: the machine she owns, and the subscriptions she chooses to give it. Every rung change is a consent gate. Nothing leaves her hardware without her agreeing to that specific crossing.
08
What she actually gets back
Nobody asks Maya for her size again. The renewals get chased without her. The upsells stop arriving. Thursday stays free. She is not managing an agent. She handed something over and it stopped being her problem.
The point was never the data. The point was the evening she got back.
Underneath. Time back, money saved, and offers that fit because they were built from what she chose to share rather than from what someone guessed.
Her agent is as powerful as the machine she owns
Not as powerful as the plan she bought. There is no premium tier here. The two dials are the hardware she already has and the subscriptions she chooses to hand it, and both are hers.
iPhone
Everyday work, always with her, always on.
Where most of her agent's day happens.
MacBook Pro
The sweet spot: serious local reasoning on a machine she already carries.
Most people's agent should live here.
🤫 Puppy One
A supercomputer at home, for work that would flatten a laptop.
For people whose agent earns its keep.
The garage grid
in designBurst capacity, owned by neighbours rather than rented from a cloud.
In design.
Deliberately relative rather than benchmarked. We have not published throughput numbers for these, so we are not going to imply we have.
Nobody subscribes for free
A party that wants something from Maya has to give her something for it. Not eventually, not in aggregate. At the moment of the grant.
- A discount or a price she would not otherwise get
- Time she does not have to spend: a form she never sees, a queue she skips
- A service that is actually right the first time because it was built on what she chose to share
- Money, where the value is plain and the exchange is honest
If a subscriber offers nothing, the honest answer is no, and the agent says it without her having to.
Free, and free for good.