I Built an AI Version of Myself. Here’s How.
What you’re reading right now was written by it.
People have said some version of “I love how your brain works” to me more times than I can count. I never know what to say to that. I assume everyone thinks the way I do.
They don’t. But I can’t see the difference from the inside.
So I built an AI version of myself to find out. This article is part of the build — drafted by the same system it’s describing, reading my own persona files as it writes.
Here’s what I built and how it works.
It started with four years of chat history
So I handed Claude over four years of my own conversations. Hundreds of chats about design systems, Figma workflows, AI strategy, client work, thought leadership, personal projects. The unfiltered record of how I actually think when I’m not writing for an audience.
The goal was to see what other people apparently already see. What came back was surprisingly specific.
It identified that I consistently build arguments in a pattern: observation, diagnosis, reframe, practical application. I didn’t know I did that. But it’s in almost every article I’ve written.
It caught that my analogies are strategic, not decorative. The bicycle for atomic design. The assembly line for backwards workflows. The garden for design system roles. Each one built for a specific audience. I knew I used analogies. I didn’t realize how systematic it was.
It flagged that I’m blunt in working contexts but modulate when writing for a public audience. I adjust more than I realized. It also surfaced a working lexicon I’d never named. Recurring terms and framings I return to across different contexts. Once you see that, you can be intentional about it.
That analysis became the foundation for everything that followed.
The file architecture
The persona isn’t one document. It’s a system of markdown files, each with a specific scope:
Bio. Career arc, education, current role. The facts.
Philosophy. How I think. Systems over one-offs. Execution over exploration. Structure enables creativity at scale.
Expertise. What I know deeply vs. broadly. Organized by depth, not keyword.
Thought Leadership. Core POVs with the actual quotes and framings I use. Positions developed through writing and conversation, not taglines.
Working With Me. What I value, what slows me down, how I communicate.
FAQ. If someone asked me this at a conference, what would I actually say?
Projects. Case studies. The decisions, the constraints, what I’d do differently.
Tools. What I use and how. The actual workflows, not a list of logos.
Content Library. Published articles indexed by theme, mapped back to core POVs.
The separation is deliberate. Bio doesn’t contain opinions. Philosophy doesn’t contain project details. Thought leadership doesn’t repeat the bio. Same principle as component architecture: single responsibility, clear boundaries, composable parts.
It’s a living system
This isn’t a project I finished. It’s a system I maintain.
Every time I publish an article, it gets cataloged with themes mapped back to my core POVs. When my thinking evolves, the philosophy and thought leadership files get updated. When I finish a project, the case study gets added.
The workflow: drop a new article into the content folder, tell Claude to catalog it. It reads the piece, adds it to the index, identifies themes, and flags if it introduces a POV I haven’t captured yet. Structured process. AI doing the tedious part.
The persona gets more accurate over time because the source material gets richer. The system compounds.
And this article you’re reading right now is being drafted inside the same system, by the same AI, reading the same persona files. The machinery is making the thing that describes the machinery.
What comes next
I’m not done building. And I’m not ready to share everything about where this is going.
But I’ve been writing that designers need to explain their taste well enough that it survives them. This is me doing that. Starting with myself, in public, one file at a time.
More soon.
I write about design systems, AI, and where the industry is actually heading. If this landed, follow me on LinkedIn. What’s your take? Drop it in the comments.