How to Give Your AI Context So It Actually Knows You

You open a new chat and type the same few sentences about yourself you typed yesterday: what you do, who you're writing for, the way you like things worded. The model nods along, does decent work, and forgets all of it the moment you close the tab. Tomorrow you open another window and do the whole thing again.

That's the quiet tax on working with AI. Every new conversation and every new tool starts you over from zero, and the thing is genuinely brilliant, and it still has no idea who you are.

Giving your AI context means handing it the durable facts about you (your role, your voice, what you're building, the people who matter) once, in a form you can reuse instead of rebuilding the whole picture from scratch every session. Done well, it gives the model a foundation to stand on, and the generic assistant slowly starts to sound like it actually knows you.

Here's how to get there.

Why your AI sounds generic

The model isn't the problem, which is the part most people get wrong.

Think about what it's actually doing when you hand it nothing to work with. It writes for the average of everyone who ever typed a request like yours, and the average, by definition, is generic. The output isn't bad so much as built for no one in particular, and right now, as far as the model is concerned, you are no one in particular. A sharper prompt won't change that — it's the difference between giving your AI instructions and giving it direction.

Picture hiring a sharp new assistant on a Monday morning. Great résumé, quick, eager to please, and completely in the dark about your clients, your standards, and the way you like a sentence to land. Every task needs a full briefing before it goes anywhere, and it stays that way every session until you change what the model knows walking in.

What "context" actually means

Context is the stable stuff about you, the parts that barely change from one Tuesday to the next.

It's who you are and what you stand for, how you communicate and the words you'd never be caught using, what you're working on right now, the stories you always come back to, and the people in your orbit and how they fit together.

Notice what's not on that list. Your chat history doesn't count, because the model mostly forgets it, and the built-in memory features that promise to fix that stay locked inside one tool. You can make ChatGPT remember you, and it still won't follow you to Claude or the next app you open. Neither does one magic prompt you found on the internet. Real context is the small, sturdy picture of you that any tool can read, the part of you that holds still long enough to be worth writing down. The thing is, most people never give their AI any of it. They hand it a task, hope for the best, and then wonder why it feels like talking to a stranger.

How to give your AI context

Start small, because you don't need all of it on day one.

1. Write down who you are. A short page on what you do, what you care about, and how you want to come across — your personal constitution, really. This one does the most work, so start here.
2. Capture your voice. A few real samples of how you write, plus the rules you follow: short sentences, no corporate filler, whatever happens to be true for you.
3. Note your situation. What you're focused on this season, including the goals, the constraints, and the things currently in motion.
4. Add what recurs. The stories you tell more than once, the people you mention, and the history that keeps surfacing in your work.
5. Keep it reusable. Put it somewhere you can grab fast, then drop it in at the start of a session or wire it into the tools that support it, like ChatGPT's custom instructions or Claude's project instructions — and here's how to write those custom instructions well.

Each of those is a context anchor: one structured, reusable piece of who you are, written once and used everywhere. Build a few of them, and the model starts working from a real foundation instead of a guess. It's the most durable of the handful of ways to give AI context, because it's the one you own and the one that travels.

The part nobody mentions

Here's the catch. Context goes stale.

You finish the project, you move cities, your priorities shift and your voice sharpens and the people in the picture quietly change, and the document you wrote with such care in January is subtly wrong by June. A confidently wrong AI, it turns out, is worse than a blank one.

So whatever you build has to stay alive. It needs to be something you revisit and keep current, not a file you bury in a folder and forget you ever made. The goal was never a one-time setup; it's a living foundation that shifts as your life does.

Where RUMO fits

I built RUMO because I was doing all of this by hand, and it was a mess.

I had pieces of myself scattered across documents, half-finished prompts, and notes I could never find when I actually needed them, and every new tool I opened wanted me to start the whole thing over. So I built the system I wished already existed: six context anchors you create once and keep current, covering who you are, how you write, what's happening now, your stories, your timeline, and the people who matter. They live in one place, and they drop into whatever AI you're using that day.

That's the whole idea, really. Give the model a place to start, so it stops guessing.

Start with one

Go back to that blank new chat, the one that knows nothing about you. Now imagine it opens already knowing how you think and how you sound, so the first answer reads like it came from someone who has worked beside you for years.

You don't get there with a better prompt. You get there by giving the model something to stand on, and you can build the first piece this afternoon.

So start with one anchor. The free one.