I pointed my personal AI memory system at my own data. What came back was more honest than I expected.
I've been building something called stonerOS—a personal AI memory system that sits on top of Claude. It tracks my career history, preferences, projects, and working patterns across sessions. But until this week, it only knew what I told it.
So I tried something different: what if it could learn from my actual behavior instead?
Your Mac already has access to most of your iPhone data through iCloud—browsing history, photos, app usage, reminders, voice memos. All SQLite databases. I built a set of Python scripts with Claude to pull patterns from each one and feed them into stonerOS's routing pipeline. Everything runs locally—no data leaves the machine.
Here's what came back.
I'm a Night Builder
My app usage data shows peak activity between 1am and 3am. Not doomscrolling—VS Code. 91 hours of active use in 90 days, nearly double the next app (Perplexity at 52 hours). Development represents 48% of my total tracked screen time.
I've always known I'm a night owl. But there's a difference between knowing that and seeing it in the data—Screen Time peaks at 1–3am (VS Code), Safari browsing surges again at 11pm, Messages cluster in the 1–4pm afternoon window. After 10pm it's all building again.
This isn't a habit to fix—it's when my brain actually works. Deep work goes late night. Meetings go in the afternoon. Lean into the rhythm instead of fighting it.
The Hobby That's Actually a Skill
My browsing data categorized 68% of my web activity as "Other." When I actually looked at what that meant: ~780 visits to collectibles marketplaces—eBay, Whatnot, PopMart, specialty jewelry sites. Nearly as much as the ~1,091 social media visits in the same period.
I collect designer toys and vinyl figures. I knew I spent time on this, but the data showed how much of it is research—not just browsing. I have an eBay scraper that tracks market prices into a SQLite database. I'm watching wholesale and secondary markets. It's systematic.
Turns out my hobbies and professional work use the same muscles—market analysis, data pipelines, pattern recognition. Just applied to different markets.
I'm Narrower Than I Think
I'm a builder in deep-focus mode. My geographic footprint is essentially two places: my apartment in Brooklyn and my family's place in rural Pennsylvania. My communication is concentrated in a tight inner circle—a handful of people I talk to daily, deeply, and consistently. My photos are 78% from home: plants, my space, my dog Theodore.
None of this was news to me. But seeing all of it together, quantified, I noticed something I hadn't named before: narrow inputs, deep engagement, high consistency. Same few apps, same few people, same few places. The pattern doesn't change—only the surface it shows up on.
This is how I operate. It builds things well. It doesn't network or explore by accident—both require intentional effort.
My Old Systems Are Time Capsules
My Apple Notes—86 of them—haven't been touched since November 2025. They contain resume drafts, job search prompts, design philosophy statements, and shower thoughts from a pre-AI version of me.
Notes from 2023 about design philosophy—observer vs. dictator, why-before-what thinking. All ideas I still believe in, written in a different role, at a different company, with different tools. A snapshot of how I thought before I had a system to build on those thoughts.
The thinking was always there. What changed is a system that builds on it instead of letting it collect dust in a notes app.
The Meta-Lesson
Honestly, none of this was a surprise. I already knew all of it. I knew I was a night builder. I knew my focus was narrow. I knew my hobbies used the same muscles as my work.
But knowing something and seeing it in data are different. "I think I'm a night person" is a vibe. "My peak cognitive window is 11pm–3am, backed by screen time, browsing history, and messaging patterns" is something I can actually schedule around.
All of this was already on my phone. I just never looked at it this way before.