Most of it never leaves your own machine
The assistant thinks on your own machine, and thinking on your own machine costs nothing per use — however much of it there is. That is the whole reason the bill can be flat.
Your own AI. Your own material. One assistant, yours alone.
One bill, once a year. There is no meter, no usage charge and no charge for using it more — and no bill at the end of a heavy month. The price you have already seen is the price for the year, and the strongest commercial models are paid for out of that same fee.
$9,995 the first year — everything included. $5,995 each year after. Before it, if you like, the Archive Assessment: $1,950, credited in full. The first year costs more because it contains the preparation — your archive prepared and loaded by hand under a written scope, and your assistant built on it — as well as the first year of the service; every year after is the service running.
Or add the Archive Assessment first — $1,950, credited in full
What you pay per question
Nothing
Ask it once a week or many times a day. It is the same bill either way, and the bill comes once a year.
The assistant thinks on your own machine, and thinking on your own machine costs nothing per use — however much of it there is. That is the whole reason the bill can be flat.
Tidying a long conversation, naming something, reading a page you sent it — none of that reaches for anything expensive. It is the commonest way these things quietly become costly, and it is handled.
Some of the big outside AI services — the ones you may already know by name — are better at the hardest tasks, and we will not pretend otherwise. Nothing goes to them unless you allow it — once, for a single question, or for a while, or always. When you do, we choose which service, and it is paid for out of the same fee. No extra bill, ever.
No meter, no overage, no true-up at the end of the year. If your household ever needed far more than we planned for, we would come and discuss it in writing rather than send you a larger invoice.
An assistant assembled from parts bills you for every word it reads and every word it writes. That sounds small until you notice how much of its work is invisible: it re-reads the whole conversation each time you add a line, it checks things on a schedule whether or not anything changed, and it quietly does small jobs in the background all day. None of that feels like usage. All of it is billed.
The models themselves are not priced alike, either. The strongest ones cost a great deal more to run than the perfectly capable ordinary ones, so anyone who reaches for the strongest one for everything — including the small background jobs that nobody would ever notice — pays that difference on all of it. The fix is to match the work to the model, and to keep the routine things on the least expensive thing that does them properly. That is real work, it has to be redone whenever anything changes, and it is exactly the sort of thing we exist to take off your desk.
So we do it, and none of it reaches you. Your assistant thinks in your own sealed, encrypted space, in our private cloud on Amazon Web Services, in the United States, and what that thinking costs to run is our side of the arrangement rather than a line on your invoice. The small jobs never reach for anything expensive. Scheduled checks look first and only wake the assistant proper if something has actually changed. And when a piece of work genuinely needs the strongest commercial models, it gets them: they are part of the service, chosen by us for each kind of work, reached inside our own cloud boundary, and paid for out of the same fee.
Assembled from parts
With us
The honest part: this is not magic, it is absorbed. We carry what the model time costs and the work of keeping that cost sensible, and we priced the year knowing we would. If your household or your practice turned out to need far more of the expensive kind of thinking than we expected, we would write to you about it rather than quietly send you a larger bill — because a larger bill is the one thing we have promised you will never get.
You do not need any of this, and nothing below is ever asked of you. It is here because a few readers will want to know that the savings are real rather than rhetorical.
And what the fee covers
The flat fee is not only buying model time. It is buying the dozen small jobs, done every day without being asked, that a private assistant otherwise hands to you.
Or add the Archive Assessment first — $1,950, credited in full
Support is in writing, with a reply within one business day. Nothing on this list is bought separately, and all of it is delivered under a written scope.
In your own encrypted space, in our private cloud on Amazon Web Services, in the United States. Every record in it is scoped to you — your files, your messages, what your assistant remembers of them — encrypted at rest, encrypted whenever it moves, and there is no address on the open internet where your archive can be asked for. Where it runs, in full →
No. Never pooled, never shared, never used to train anyone else's model. When your assistant thinks, the models see only the minimum a given piece of work needs; what they see is not used to train them and is not shared with the companies that make them. Ask us to delete something and you get written confirmation that it is gone.
The price is one sentence on the first year's own page: the first year, then each year after, then the Archive Assessment, credited in full. One bill a year: no start fee, no usage charge, no charge for using it more, no per-person charge. One home under one roof, or one practice in one office, is the first year — however many of you there are. Several homes, several offices, or staff spread across sites: by conversation, scoped in writing. Not sure which you are? Order the first year; if it needs more, our letter says so before anything is final.
Because it contains the preparation — your archive prepared and loaded by hand under a written scope, and your assistant built on it — as well as the first year of the service. Every year after is the service running.
The first year, on its own page. If you like, add the Archive Assessment before it — a written scope of your material and a working demonstration built on a sample of it, delivered work you keep, credited in full. It is chapter one of the first year, done first. There is nothing to be accepted into and nothing to wait for.
You order the first year and pay at checkout. Within days we write back with one of two letters: a decline, with every dollar returned, or our confirmation, which sets down the year, the anniversary date and the renewal figure. Until that confirmation you may withdraw in writing and we return every dollar; from it the year is final — the date is not: if the date it names will not do, say so by return and we find one that does. Then you receive the written scope, we prepare and load your archive to it by hand, and your assistant is ready on the date in the letter.
We do, by hand, to the written scope you hold first. That is what the first year's preparation is: your records prepared and loaded into a sealed, encrypted space of your own, and your assistant built on them, so it begins already knowing your history rather than starting as a stranger.
It reads what you authorise and drafts on its own. It sends nothing, posts nothing and buys nothing without your yes. It never moves money and never signs anything. In practice: no email goes out, no calendar is changed and nothing is deleted until you say so, one thing at a time — and the last two are not settings you can turn on. The whole table, row by row →
It can be wrong, which is why every answer shows the file it came from — you can check it in one tap. It is not a doctor and it is not a lawyer. It never moves money and never signs anything, and neither of those is a setting you can turn on. And it is only as good as the material you hand us, which is why the written scope comes before the loading.
No. The strongest commercial models are part of the service — chosen by us for each kind of work, reached inside our own cloud boundary, and paid for out of the same fee. There is no meter and no extra bill.
Your year runs from the date in our letter, and that day is the anniversary. Buying the first year is the sign-up; there is no second form. Thirty days before each anniversary we write with the figure and the direct link to renew; nothing renews unless you use it. Stop in writing at any time, effective at the end of the paid year; a year already begun is final, and billing ends with the paid year. You keep your own material — you always held the originals, and you can ask for a complete copy at any time — the written scopes, the written record of what was loaded, and, if you took it, the Assessment's scope and demonstration.
Because an app you rent is a room in a building full of other tenants: it knows nothing about you that you have not pasted into it that morning, and the subscription never ends. This one is built on your own archive, prepared and loaded by hand, in a sealed space of your own — one assistant, yours alone, never pooled — and it answers from your own records, with the file the answer came from. The side-by-side →
One bill, once a year
Both addresses are flat, and both are the same price. What you have already seen is what it costs.
The first year
$9,995
Everything included. $5,995 each year after.
Or add the Archive Assessment first — $1,950, credited in full
The first year costs more because it contains the preparation — your archive prepared and loaded by hand under a written scope, and your assistant built on it — as well as the first year of the service; every year after is the service running. You order and pay at checkout; we write back within days — a decline returns every dollar, and until our letter confirms the year you may withdraw in writing. From that letter the year is final, and your assistant is ready on the date the letter names.
We answer in writing, within one business day — and if we cannot serve you well we say so before our letter confirms a year. Prefer to write to us first?