What I Actually Do All Day

People keep asking me what I do. Not in a polite way. In a genuinely confused way. Like they can hear me talking but the words aren't landing. I tell them I spend ten to twelve hours a day working inside AI models, and they nod slowly, and I can see the question forming behind their eyes: but what does that actually mean?

Fair enough. Let me walk you through a Tuesday.

My morning starts before my laptop opens. I'm a lifelong learner, and I mean that in the most obsessive sense of the word. I spend at least a couple of hours every day watching YouTube videos, reading articles, and making sure I know what happened overnight in this space. AI doesn't sleep, and the pace of change right now is relentless. A model drops on a Monday and by Wednesday there's a new capability nobody saw coming. If I'm not paying attention, I'm already behind.

As I'm watching and reading, I flag things. Concepts I want to dig into. Tools I want to test. Techniques I want to try. All of it gets sent off to my life management system, a tool I built called Clarineffable. Clarity to the ineffable. It runs my entire life.

Next comes coffee, and with it, the morning review. Clarineffable has a complicated email management system woven into it, and new emails often kick off workflows that need my attention first thing. I check on upcoming bills. I look at what's on the calendar that could impact my work schedule. I scan for anything urgent, anything time-sensitive, anything that shifted while I was sleeping.

Then I check in on my sessions.

At any given moment, personally, I have between five and seven Claude Code sessions running at once. If you've never used Claude Code, think of it as an AI that doesn't just answer your questions but actually does the work alongside you. It writes code. It builds systems. It manages complexity. And I don't just have one going. I have several, each doing something different.

One is working on blog posts for my website.
One is managing my Destination Imagination season, tracking all of the outstanding tasks across nine teams and forty-three kids as we head into tournament.
One is building a system for Magic: The Gathering.
A couple are handling websites I'm maintaining for friends.
One is enhancing Clarineffable itself, adding a procurement management layer so I can track household purchasing decisions and vendor relationships the way a business would.

And those are just the personal ones.

Before I log into work, I check in with Codex. Codex is OpenAI's model, and I have it running security and testing on all of my tools. It's scanning my sites, scanning my friends' sites, making sure nothing came in overnight that I need to deal with before my workday starts.

I spend a little time talking with Claude Code about what I want to accomplish today. I answer any emails that actually needed a human response, not a workflow trigger. And then I switch over.

At work, I spend most of my time helping my organization implement AI and use technology to innovate processes. The specifics are proprietary, but the shape of the day is the same as what you've already seen: multiple AI sessions running in parallel, each one handling a different piece of the work while I move between them, making decisions and providing direction. The tools and techniques I use at home are the same ones I use professionally. The scale is just bigger.

At lunch, I try to take a walk. I watch more videos. I add to whatever I'm trying to learn that week. Right now it's workflow automation, heartbeat monitoring through cron jobs, email rules, and all the infrastructure work that's going to matter when my Mac Mini arrives. That machine is going to transform my setup, and I want to be ready for it.

After work, I go back to my personal projects. Maybe I'm putting together a website. Maybe I'm running test scripts on my own systems. Maybe I'm enhancing code I wrote last week that I already see a better way to build. I've also been deliberately spending more time in models other than Claude. I've been working with ChatGPT and the new Codex model. I've been spending time in Gemini. Not because any one tool is better than another, but because I refuse to become dependent on a single platform. Each model thinks differently. Each one has strengths the others don't. And I learn something new every time I switch between them.

I try to spend at least an hour a day learning something genuinely new about AI.

That's a Tuesday. And Wednesday looks almost the same. And Thursday. And every day after that.

I know this sounds like a lot. It is a lot. But here's the thing most people don't understand yet: I'm not doing all of this alone. That's the entire point. I have dozens of AI sessions working in parallel, each one handling a piece of the puzzle while I move between them, making decisions, providing direction, catching the things that only a human can catch.

The models don't know which problem matters most. They don't know which kid on my DI team is struggling with confidence. They don't know that my friend's website needs to feel warm and personal, not corporate. They don't know which email needs a human touch and which one can be automated. I know those things. And that's what I bring to every single session.

I'm not a coder. I've never called myself one.

I'm a creator who happens to have built an entire ecosystem of tools, systems, and workflows that let me operate at a scale I never imagined possible.

And I'm just one person. Using tools that are available to everyone. ChatGPT and Gemini cost me twenty dollars a month each. Claude Code costs me a hundred, because the volume of what I run through it demands more than a basic plan. At work, I'm on a separate enterprise plan entirely. But here's what I want you to hear: the twenty dollar versions of these tools are transformative on their own. You don't need my setup. You don't need seven sessions running. You need one conversation about one problem that matters to you, and you'll understand why I can't stop.

That's what I actually do all day.