Sometimes AI Doesn't Build Confidence. It Reflects It.

A few weeks ago I spent some time helping friends build personal portfolio websites. I love this kind of project. It's practical, it has a clear finish line, and getting to use AI to do something genuinely useful for people I care about is the best kind of afternoon.

Except the websites weren't really the project.

I didn't realize that until I was sitting with the third friend, watching her read her own profile back to herself for the first time. She stopped mid-paragraph. Her face changed. And she said, quietly, almost to herself: "That actually sounds like me."

Followed a moment later by: "I had no idea I'd done all of that."

The first friend had smiled when she said it. The second had laughed — a surprised, slightly disbelieving laugh. The third got a little emotional. But all three said some version of the same two sentences, in the same order, with the same pause in between.

That pause is what I've been thinking about ever since.

What we actually did

The typical approach to building a portfolio or bio is to hand someone a blank page and ask them to write about themselves. Which, if you've ever tried it, you know is a specific kind of uncomfortable. Most people either undersell dramatically or freeze completely.

I tried something different. Instead of starting with the blank page, I asked each friend to open ChatGPT and ask it a question they'd never thought to ask before:

"Based on everything you know about me — my career, my conversations, my work — what have I actually accomplished? What strengths do you see showing up consistently? What themes run through what I've built and how I show up for people?"

Then we layered in their resume, their LinkedIn, examples of their work. AI organized the raw material into a coherent story. Claude shaped it into the voice and format for a portfolio page.

The process worked well. The websites came out well. But that's not what stayed with me.

What I didn't expect

What I didn't expect was the moment when each person heard their own story read back to them — assembled by something that had no stake in flattering them, no knowledge of their bad days, no memory of the projects that almost didn't work out.

AI doesn't know about the presentation you bombed. It doesn't remember the quarter where nothing landed. It doesn't carry the weight of every revision, every difficult stakeholder, every late night you spent wondering if you'd made the right call.

It just looks at what's there. It notices patterns. It connects things across time. It names strengths that have been present so long they've become invisible.

That's a very different lens than the one most of us use when we look at ourselves.

Imposter syndrome works the way it does because we're comparing our inside view to everyone else's outside view. We have access to all the uncertainty and doubt and effort. They only see the result. It's not a fair comparison, and yet it's the one we make constantly.

What this made me think about

I've spent a lot of time thinking about what AI is actually for. I build systems that reduce mental load, automate documentation, help families stay organized, help teenagers find their footing. Those feel like obvious, practical applications.

This one surprised me, because it was none of those things. It wasn't about efficiency. It wasn't about saving time. It was about perspective.

There's something quietly powerful about having your story reflected back to you by something that has no reason to minimize it. No awkwardness about seeming like you're bragging. No pull toward false modesty. Just: here is what's actually there. Here is what the pattern looks like from the outside.

I think about the people I know — especially women in technology — who are genuinely exceptional at what they do and have no idea. Who are so focused on the next challenge that they've never stopped to look at the body of work they've already built. Who would never say out loud what AI said about them without flinching.

Hearing it from AI somehow made it easier to believe.

The question I'm sitting with

I started this project thinking I was helping people build websites. I ended it thinking about how rarely we give ourselves credit for the things we've actually done — and how much of that is just the cost of being the one who lived through it.

We remember the effort. Everyone else remembers the impact.

AI, it turns out, remembers the impact too. And on the right day, with the right question, it can hand that back to you in a way that's hard to argue with.

I don't think AI knows us better than we know ourselves. But I do think it can sometimes see us more clearly — precisely because it doesn't know what it was like when things were hard.

That's not a small thing.