If you start reading about how to make an assistant know your own material, you will quickly meet two pieces of jargon: retrieval and fine-tuning. They are usually presented as rivals, and the argument between them is usually conducted by people who enjoy the argument.
You do not need the jargon. You need to understand two different things a system can do with your papers, because they fail in different ways — and because one of them is routinely sold as something it is not. We will say where we stand as we go.
The first way: looking it up
The first method keeps your material as material. When you ask a question, the system searches your documents, finds the passages that seem relevant, and reads them before answering.
This is closer to a very fast assistant with a very good filing system than to anything mysterious. The model itself has not changed. It has simply been handed the right pages at the right moment.
What it is good at. Facts that must be exactly right. A figure, a date, a clause. Because the system is reading the actual document, it can show you the source, and you can check it. It also updates instantly: add a document this morning and it can be used this afternoon.
Where it struggles. Questions that are not about one passage. “Are we drifting over budget on this project?” is not answered by finding the right page, because no page says it. The answer lives across hundreds of documents, and a single lookup is poorly suited to that shape of question.
The second way: learning it
The second thing you want is harder to name and easy to recognise. An assistant that has learned your world knows your vocabulary without being told, knows which supplier you mean when you use a nickname, knows the shape of your year, and writes in a way that sounds like your practice rather than like a press release.
The jargon answer is fine-tuning: adjusting the model’s own machinery on your material, so that your world becomes part of how it thinks rather than something it consults. It sounds like the serious option, which is why it is the one most often sold. For a private archive, it is mostly the wrong tool.
Why. A model whose machinery has absorbed a thousand invoices has an excellent sense of what your invoices are like, and should not be trusted to recite the total of invoice 407 from memory. Ask it to, and it may produce a plausible number. That is the failure worth understanding, because it is confident and wrong rather than blank. And when a fact changes — a supplier, an address, a decision reversed — machinery cannot be corrected the way a record can.
There is a plainer route to the same familiarity: reading, and organised memory. Your archive is loaded and organised so the assistant learns your vocabulary, your habits and your history the way a careful new colleague does — by reading, and by keeping a memory of what it has learned about you. A memory has two advantages machinery does not: you can look at it, and you can change it. Tell it the nickname retired, and it is retired; ask for something to be forgotten, and it is forgotten.
Why the argument is a false one
Set out like that, the answer is obvious: you want both effects, each produced the way it is produced well.
Learning gives the assistant a sense of your world — the vocabulary, the habits, the judgement about what matters. Looking up gives it the facts, with the source attached, so that any specific claim can be checked against the document it came from.
A system built on only the first is a stranger with a good search box. A system built on only the second is a colleague with a confident memory and no filing cabinet, which is a genuinely dangerous combination for anything involving numbers.
Since this is our page, here is how we do it, plainly. No training run happens on your material: nothing you give us adjusts a model’s machinery, and nothing you give us is ever used to train anyone else’s model. Your archive is loaded and organised, by hand, under a written scope, so your assistant learns your world by reading it — and what it remembers about you is yours to correct or delete, honoured and confirmed in writing. Every specific fact still comes the first way: looked up, with the document it came from attached.
What this means for you
Three practical consequences, if you are evaluating anything in this category.
Ask what happens to a number. If you ask for a figure, does the answer come with the document it came from? If it cannot show you, treat every number it gives you as a draft.
Ask how quickly new material becomes usable. A document you add this morning should be usable this afternoon, not after the next rebuild. An assistant that is always slightly out of date is worse than it sounds when the subject is your own affairs.
Ask whose boundary the reading happens in. Both methods require something to read all of your material, and that is the privacy question. It is not settled by which method is used, nor really by where a computer sits — it is settled by whose boundary the reading happens inside, and what leaves it. In our case: your material sits in your own sealed space in our private cloud on Amazon Web Services, scoped to you and nobody else — never pooled, never shared, never used to train anyone else’s model.
The part that is the same either way
Both methods depend entirely on the state of your material. Neither one rescues an archive full of contradictions, duplicates and unreadable scans — looking up will find the wrong version, and learning will learn the wrong version.
Which is why the first useful step is not choosing a method. It is finding out what you actually have.
What can and cannot be loaded →
Get a written scope of your own material →
Your own AI. Your own material. One assistant, yours alone.
$9,995 the first year — everything included. $5,995 each year after. 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
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 it arrives on the date the letter names. Prefer to write to us first?