My Client Said AI Knows Him Better Than Most People Do. He Was Right.

A client reached out after building his first context profile. He sounded half amazed, half unsettled.

He spent time reading through it. The document described how he makes decisions, how he batches things, when he does his best work. His reaction: it probably knew him better than most people do.

Here’s what happened. He had taken something I show people early on and ran with it. He gathered his own work data and asked the AI one simple question: “Based on all of this, write a detailed profile of who I am, how I work, and how I make decisions.”

The result shook him.

The Mirror Effect

Most people use AI like a search engine. You ask it a question, and it gives you an answer. There’s a clear boundary between you and the tool. You’re the one with all the context.

But when you feed AI your actual data (months of emails, meeting transcripts, calendar patterns), something different happens. It doesn’t just know facts about you. It starts to see patterns. Patterns you might not even see yourself.

The profile nailed things people closest to him probably couldn’t have articulated. Not because the AI is smarter than a human. But because it processed all the data without the filters we put on ourselves. We tell ourselves stories about how we work. The data tells a different story sometimes.

That’s what caught him off guard. Not the speed. Not the writing quality. The accuracy.

Why This Matters for Agents

This kind of profile is the single most important thing you can build if you want AI agents that actually work for you.

Think about it this way. When you hire an executive assistant on their first day, they’re basically useless. Not because they’re incompetent. Because they don’t know you. They don’t know your priorities, your communication style, which meetings matter and which ones are a waste of time. They don’t know that you hate mornings or that you make your worst decisions after 4 PM.

By month six, that same EA is running your life. The only thing that changed is context.

AI agents work the same way. A generic chatbot gives you generic answers. An agent with your context profile gives you answers that fit your actual life.

I have a dozen or so agents running now. My morning briefing agent, my meeting prep agent, my email drafter, my scheduling assistant. All of them read from the same shared context. Same priorities. Same communication preferences. Same list of people I know.

That shared context is what makes the whole system feel like one teammate instead of a pile of separate tools.

The Pilot Analogy

One of my clients put it in a way I keep coming back to. He said most people’s days are like a pilot constantly taking off and landing. That’s the stressful part of flying. You take off for deep work, land for a meeting, take off again for admin, land for a phone call. By 3 PM your brain is fried.

With agents handling the prep work, follow-ups, and admin, you take off once at the beginning of the day and you cruise. You land once at the end. That single shift from constant context switching to sustained focus is where the real value lives.

But it only works if the agents have enough context to handle things without you. Otherwise you’re still the one doing the mental takeoffs and landings. You’re just doing them through a chatbot instead of your inbox.

How to Build Your Profile

The simplest version takes about an hour:

1. Export your data. Grab a few months of sent emails, your calendar for the last quarter, and any meeting transcripts you have.
2. Upload to ChatGPT or Claude. Use a long-context model so it can process everything at once.
3. Ask the right question. “Based on all of this data, write a detailed profile of me. How do I make decisions? When am I most productive? What are my communication patterns? What drains me? What energizes me? What are my blind spots?”
4. Read it and refine. Some things will be off. Fix them. Add things it missed. The first version is about 80% right. You make it better over time.
5. Share it with your agents. This becomes the context file every agent reads before doing anything.

The key is that this document isn’t static. I update mine every few weeks as things change. New projects, shifted priorities, lessons learned. It’s a living document that keeps getting more accurate.

Do This This Week

You don’t need a year of data. Export 30 days of sent email and last month’s calendar. That’s enough to get a first mirror.

When you read the draft, mark 3 lines that feel true and 3 that feel off. Fix the off ones yourself. Then use that profile on one weekly plan, not on every tool you own.

If the profile says you do your best work in a certain window, protect that window. If it says you crash after back-to-back calls, put recovery on the calendar instead of pretending you don’t need it.

It Gets Better Over Time

My Digital Chief of Staff system didn’t happen overnight. It was an accumulation of small improvements over about a year. I’d add one agent, let it run, tune it, then add the next one.

But the context profile was the foundation. Everything else was built on top of it. Without it, every agent is just guessing. With it, they actually know what you care about.

His reaction was the same one I see with almost everyone who tries this. A mix of “this is incredible” and “this is kind of weird.” The weirdness fades. The usefulness doesn’t.

Build the profile. Read it. Fix it. Share it. Then watch everything else start working better.

If you want a lighter starting point, try Focus Filter. Get the noise down first so the profile has room to work.