In a past life, I ran a digital agency. Built it from my basement, crammed four of us into my living room, eventually got a real office (garbage bags on the windows for three months until we could afford curtains). Grew it into a real company with real clients and real people depending on it. Sales. Strategy. Client management. HR. Operations. And then the actual building: 40+ hours a week of hands-on development, stacked on top of the 40-60 hours of everything else. That was the norm for years. 80-100 hour weeks, because that's what it took.
I was great at whatever I was doing right now. Everything else went dormant.
I was always good with clients. I could walk into a meeting, nail the problem, make them feel like the only account that mattered. The people who bore the brunt were my own team. Teammates would need five minutes that turned into 45 because I'd lost track of what they were waiting on. I'd be deep in a technical build and there'd be three people outside my door needing decisions I'd forgotten I owed them. They learned to work around it. Some of them were gracious about it. Others were not.
Whatever had my focus had ALL of me. The rest of the world could wait.
This was not a productivity problem. I was productive as hell, just on the wrong things at the wrong times. I could not distribute my attention across multiple workstreams without dropping half of them. Whatever I was building got my best work, and whatever I'd set down got neglected until the damage was done.
Then AI showed up.
Technically, AI showed up twice.
The first time was ChatGPT, while I was COO at the law firm. It replaced Grammarly and Google search overnight. One-off advice. Dinner ideas. Help drafting emails. Cleaning up writing. It was a game-changer at the time, but it was still a tool I reached for occasionally, not one I lived inside.
The second time was Gemini, after I joined what's now Ateko. And this wasn't gradual. I went from 17 prompts in March 2025 to 825 in April (at time of writing, those numbers have kept climbing). That's not "trying a new tool." That's finding something that fits how your brain works and going all in before you can articulate why.
Within two months I was typing hundreds of thousands of words a month into AI conversations. Not copy-pasting code or dumping documents. Thinking out loud, across multiple projects, for hours at a time. Not all of it was gold, but the volume was the point. I was externalizing my thinking into a tool that could hold context while my brain bounced to the next thing.
And I wasn't scattered. I was conducting.
What changed when I started using AI every day?
I went from one project at a time to four in a single day. The AI held context while I bounced between them.
Before AI (the agency years): I had one thread running at a time. If I was building a website, proposals piled up. If I was doing sales, the client work stalled. Context-switching wasn't just unpleasant; it was destructive. I'd lose 30 minutes getting back into the headspace of whatever I'd left behind. So I didn't switch. I stayed on one thing until it was done or I was forced to move. Then the other thing was behind.
The early Gemini phase (April to November 2025): I started running 3-4 Gemini windows simultaneously. One for an RFP I was responding to. One for a Chrome extension I was building. One for Commerce Cloud technical work. One for coaching a colleague's blog post.
Each window held context. I didn't have to re-explain where I was. I could bounce from the RFP to the Chrome extension when my patience ran out on a stubborn section, do 20 minutes of productive work there, and bounce back.
I wasn't losing context anymore. The AI was holding it for me.
The Cursor phase (February 2026 onward): This is where the real shift happened. Not "AI made me faster." AI let me keep 2-5 things alive at the same time.
Cursor doesn't just hold context in a conversation. It holds context in your entire codebase: rules files, working notes, project configuration, code. I could set up a project, invest an hour getting the AI oriented with the right context, planning, and direction, and then walk away. When I came back three days later, the context was still there. The rules I'd written were still active. The AI remembered what we'd decided and why.
My typical workflow now: I have 2-5 Cursor windows open, each on a different initiative. I spend time up front making sure each one has the right context and direction before I consider it "ready." Once it's ready, I bounce between them. Twenty minutes on this one, check the output on that one, give the third one a new task while I'm waiting.
On April 6th, I touched four projects in a single day: a standards framework at 9 AM, a content import tool at 10:30 AM, a security audit project at 2:30 PM, and a certification coaching app at 5 PM.
I am not meaningfully faster at any one of those tasks. I probably produce a better product in roughly the same time, because the AI catches things I'd miss and I catch things the AI gets wrong. But the real gain is that I'm getting 2-5x MORE things done. Not faster. More. That's a fundamentally different claim than "10x productivity," and it's the one I can actually back up.
Did AI actually make me more productive?
Not faster. Broader. The real gain was keeping four projects alive simultaneously instead of dropping three while I focused on one.
I want to be careful here, because the internet is full of "AI made me 10x productive" stories and most of them are bullshit. The independent research says the real productivity gain is modest, somewhere around 15-20%, and there's evidence developers believe they're faster even when they're not.
I'm not claiming 10x anything.
What I'm claiming is this: AI didn't make me faster. It made me broader.
The things I was already good at, I'm still good at. But the things I used to drop, the things I was terrible at because they didn't have my attention, those are the things that changed. I can keep four projects moving now because I don't have to hold all the context in my own head. The AI holds the context while I bring the judgment.
The best analogy I've found is this: I used to be a solo performer. Now I'm more like a conductor. The orchestra (the AI tools, the context files, the rules systems) keeps playing while I move my attention across the sections.
The Uncomfortable Part
I have ADD. Not formally diagnosed, but all the telltale signs are there. In the principal's office every day in grade primary. Struggled all through school to focus, control my impulsivity, sit still. It wasn't until I settled into doing things I actually loved that I found out I was a machine.
I know the clinical term changed years ago. ADD is how my generation learned it, and it's how I still think of it.
Whatever had my focus had all of me. Switching was painful. I'd hyperfocus on one thing and go dormant on everything else. I'd be impatient waiting for anything that took more than a few seconds. I'd start conversations in the middle because the beginning felt obvious.
I'm not going to dress this up in the language of the moment. "Neurodivergent" has become the new hustle culture badge, and I have no interest in wearing it like a personality trait. But I'll say this plainly: the specific things that made me bad at plate-spinning (the hyperfocus, the impatience, the context-switching cost) are the exact things AI tools are built to compensate for. Palantir's CEO Alex Karp told Fortune "there are basically two ways to know you have a future: vocational training in the trades, or being neurodivergent." I don't care about the label. I care about the observation.
AI holds context so I don't have to. It lets me ramble and then organizes my thoughts. It lets me switch tasks when my patience runs out without losing the thread. It gives me something to do while I'm waiting (and I have no patience for waiting).
Louis CK has this bit from 2008 called "Everything's Amazing and Nobody's Happy." He's on Conan, and he's talking about a guy on a plane:
"I was on an airplane and there was high-speed internet on the airplane. That's the newest thing that I know exists. And I'm sitting on the plane and they go, open up your laptop, you can go on the internet. And it's fast, and I'm watching YouTube clips, it's amazing. I'm in an airplane. And then it breaks down. They apologize, the internet's not working. The guy next to me goes, 'pfft, this is bullshit.' Like, how quickly the world owes him something he knew existed only ten seconds ago."
And then the kicker:
"People come back from flights and they tell you their story, and it's like a horror story... 'We didn't board for twenty minutes, and then they made us sit there on the runway...' Oh really, what happened next? Did you fly through the air, incredibly, like a bird? Did you partake in the miracle of human flight, you non-contributing zero?"
That's my relationship with AI. I'm sitting in a chair in the sky, watching a machine think for me, and I'm going "really? Can we hurry up here?" Gemini takes four seconds to respond and I've already switched to another window because I can't sit still that long.
The impatience is the feature, not the bug. It pushes me to the next window, the next project, the next conversation. AI made that productive instead of destructive.
What should you do if your brain works like this?
-
Invest the setup time. The first hour with a new AI project feels slow. You're loading context, explaining what you're building, setting up rules. That hour pays off for weeks. I have Cursor conversations that span 40 days because the initial setup was thorough enough to sustain it.
-
Don't fight the bounce. If your attention wants to shift, shift. Open the other project. Do 20 minutes there. Come back. AI holds the context you're leaving behind. This is the opposite of every productivity book I've ever read, and it works better for my brain than any of them.
-
Write things down for the AI, not for yourself. Rules files, project descriptions, working notes. You're not documenting for future-you. You're documenting for the AI that's going to pick up where you left off. The better the documentation, the faster the re-entry.
-
Start with one project and one tool. I didn't go from zero to thousands of prompts overnight. I went from zero to "huh, this is interesting" to 825 prompts in one month because I found one problem that fit. Find your one problem first.
-
Track it. I didn't know I'd typed 3 million words until I exported the data. Looking at the numbers changed how I understood what the tool was doing for me. It stopped being "I think this helps" and became "I can see exactly how this helps."
The Bigger Point
The conversation about AI and productivity is mostly being had by people who think about work the same way. Linear. Focused. One thing at a time, done well, move to the next.
If that's your brain, great. AI probably makes you 15% faster and that's still worth it.
But if your brain doesn't work that way, if you're wired to bounce, to hyperfocus, to carry five things at once and drop three, then AI isn't a productivity tool. It's a cognitive scaffolding. It doesn't change how fast you think. It changes how many things you can think about without losing the ones you're not looking at right now.
That's what it unlocked for me. Not speed. Breadth.
I still have garbage bags for curtains in some part of my life. The difference is that now, when I set something down to hyperfocus on something else, there's an AI holding the thread until I come back. The things I set down don't go dormant anymore. They just wait, with full context, for my attention to circle back around.
If your brain works anything like mine, that might be the most important thing a tool has ever done for you. It was for me.