The 8 Files That Let Any AI Start From Who You Already Are
The reason an AI feels like it doesn't know you is simple: it is reading nothing about you. Every conversation begins at zero. You spend the first ten messages re-explaining your work, your taste, and the things you will not do — and then the window closes and the next chat forgets all of it.
The fix is not a better assistant. It is a set of files. Eight of them, each holding one kind of context, each plain enough that any AI can read it. Hand them over and the AI starts from who you already are.
Here is what each file holds, and the exact moment it earns its place.
Why Split It Into Eight
You could write one long document called about me. People try this. It fails for a quiet reason: a single file mixes your principles, your projects, and your hard limits into one paragraph soup, and the AI has to guess which part matters for the question in front of it.
Splitting by purpose fixes the guessing. Your beliefs live in one file. Your hard rules live in another. Your work history lives in a third. When a task only needs two of the eight, you hand over two — not your entire inner life. Separation is not bureaucracy. It is how the right context reaches the right question.
The 8 Files, One Problem Each
MY_CANON — your principles
The problem it solves: the AI gives advice that contradicts what you actually believe. This file holds your first principles, the ideas you build on, the lines you've already settled. Ask it to help you make a decision and it argues from your values, not the internet's average opinion.
MY_PORTALS — the doors into your world
The problem it solves: the AI has no idea where you actually live online or what your front doors are. This file lists your public surfaces — your site, your channels, the places people find you — so the AI can point at the real thing instead of inventing a plausible-sounding link.
MY_ARTIFACTS — what you've made
The problem it solves: the AI keeps suggesting you "build" things that already exist. This file is the shelf of what you've already made — the products, the writing, the work that's done. Now the AI extends what's there instead of reinventing it from scratch.
MY_OPERATIONS — how you actually work
The problem it solves: the AI gives you a workflow that has nothing to do with yours. This file describes how you run things day to day — your cadence, your tools, the order you do things in. Ask for a plan and it fits the way you already operate, instead of handing you someone else's routine.
MY_REVENUE_MAP — where the value comes from
The problem it solves: the AI optimizes the wrong thing because it doesn't know what pays. This file maps how your work creates value — what matters, what's a side project, what's just for love. Now its suggestions point at what actually moves the needle, not whatever sounds impressive.
MY_RED_LINES — what it must never do
The problem it solves: the AI cheerfully crosses a line you care about. This is the file of hard limits — the private details that never leave, the names that never appear, the boundaries you set on purpose. It is the one file you would never want an AI to skip, because it's the one that says stop.
MY_ARCHIVE — what's already behind you
The problem it solves: the AI keeps dragging you back to things you've already closed. This file holds the paused, the retired, the done-and-dusted. It tells the AI what not to revive, so it stops handing you advice about a chapter you finished a year ago.
MY_FIRST_7_DISPATCHES — where to start
The problem it solves: the AI reads everything, then asks "so what do you want to do?" This file is the opening move — the first handful of concrete tasks worth doing given everything in the other seven files. It turns a pile of context into a first action, so the conversation starts working instead of warming up.
The Mistake Most People Make
The common advice is to write a short "AI persona prompt" — a paragraph that says you're a friendly marketer who likes concise answers. People paste it into every chat and feel organized.
It barely helps. A persona paragraph tells the AI a vibe, not a life. It has no projects in it, no history, no list of things you will not do. The AI reads "concise and friendly" and still has to guess everything that actually matters. A vibe is decoration. Context is the eight files. The difference shows up the moment you ask a real question instead of a generic one.
An assistant that knows your tone but not your red lines is a stranger with good manners.
You Probably Already Wrote Most of This
Eight files sounds like homework. It isn't, if you already write anything down. A year of journal entries holds your canon and half your red lines. Your project docs hold your artifacts and operations. Your notes hold portals you forgot you had. The raw material is scattered across things you've already written; the work is mostly sorting, not inventing.
That sorting is exactly what Soul Alchemy does. You paste your existing writing — articles, journal, notes — and it produces the eight files as drafts: MY_CANON, MY_PORTALS, MY_ARTIFACTS, MY_OPERATIONS, MY_REVENUE_MAP, MY_RED_LINES, MY_ARCHIVE, and MY_FIRST_7_DISPATCHES. You edit them, because no first draft of yourself is exactly right. But you're correcting a draft, not facing eight blank pages.
Once the files exist, they're yours. Plain markdown, readable by ChatGPT, Claude, Gemini, or whatever you use next. Switching tools stops meaning starting over. You bring the files; the new assistant reads them; the conversation begins where the last one left off.
Start With Two
You don't need all eight to feel the difference. Write MY_CANON and MY_RED_LINES first — what you believe, and what an AI must never do. Those two carry most of the weight. The other six make the context richer, but principles and limits are what change the answers from generic to yours. Two files today beats eight perfect files someday.
Frequently Asked Questions
What are personal AI context files?
They are plain text files, usually markdown, that describe who you are in a way an AI can read at the start of a conversation. Instead of re-explaining your work, your taste, and your rules every time, you keep them in files and hand the files over. The AI starts from your full context instead of from zero.
Why eight files instead of one big file?
One file mixes everything together and forces the AI to guess what matters. Splitting by purpose keeps each part clean: your principles in one place, your hard rules in another, your work history in a third. You can also hand over only the files a task needs, instead of dumping your whole life into every chat.
Which file matters most?
MY_RED_LINES, the file of things the AI must never do. Principles and history make answers better. Red lines stop the answers that would actually cost you something — leaking a private detail, naming the wrong person, crossing a boundary you set on purpose. The one file you would not want an AI to skip is the one that says stop.
Do I have to write all eight files by hand?
No. If you already write — a journal, essays, notes, project docs — most of the raw material exists. A tool like Soul Alchemy reads your existing writing and drafts the eight files for you, which you then edit. You are correcting a draft, not staring at eight blank pages.
Can I use these files with any AI?
That is the point of keeping them as plain markdown. ChatGPT, Claude, Gemini, or whatever you use next can all read text you paste or upload. The files belong to you, not to one platform, so switching tools does not mean re-teaching a new assistant from scratch.
Generate Your 8 Context Files From Writing You Already Have
Soul Alchemy reads your articles, journal, and notes and drafts all eight files — MY_CANON through MY_FIRST_7_DISPATCHES — that any AI can read. $99, no subscription.
Try Soul Alchemy