Private notes, made public. · 10th Jul, 2026

Hey, it’s Gabriel.

These are my private notes this week on building wealth, raising my family, and breaking generational patterns.

This week
⚡ an AI twin of me fooled every judge in a blind test
❤️ my wife just finished a confinement she never started
💰the scoreboard: one position down 50%, one up 127%, one exit
🧬 and the one-line focus rule that keeps my ADHD brain from eating itself

I tried this once before, back in February. With the models available then, the result was embarrassing. It sprinkled Singlish into serious work messages where I never would, and it wrote to my wife like a greeting card. Anyone who knew me could smell the fake in two sentences. I shelved it.

Then Fable 5 came back, and I decided the problem was never the idea. It was the model.

So I redid the entire thing.

This time my agents collected everything I have ever written or said that we could find: 211,865 messages. My WhatsApp. My Slack. Two years of conversations with AI. Even 21,990 lines of me literally speaking, transcribed from calls. Then the AI interviewed me, 31 questions, to catch the things the data got wrong about me. It caught real errors, including a story it thought was mine that actually belonged to someone I had interviewed.

From all that, it built five versions of my voice. How I talk to my team is not how I talk to AI, is not how I write in public. The old twin never knew the difference. This one does.

Then I ran the test. 22 unseen prompts. The February twin and the new twin both answered every one. Names stripped, order shuffled, and three separate AI judges scored them blind.

The new twin won 17 to 4, with one tie. All three judges.

The Honest Part: two of the three judges are from the same model family that generated the answers, so I hold the result loosely. And one rule is carved in stone: the twin never pretends to be me. Anywhere it acts, it discloses what it is. I want a tool, not an impostor.

Here’s the part that matters for you. The February version failed not because cloning yourself is a bad idea, but because it was a too-early idea. The models turned over a generation, and a shelved project became a working one in a week.

Which project did you shelve because the AI wasn’t good enough at the time, and have you checked it lately?

If you didn’t grow up Asian, confinement is the traditional first month after birth: the mother stays home under a list of rules. No washing your hair. No showering, for a month. Only certain teas and warming waters allowed.

Zoey is officially “out of confinement” today, having done none of it. Let me explain why with my favorite story.

A couple gets married. Every time the wife cooks fish, she cuts off the head first. One day the husband asks why. She says, “I don’t know. My mom always did it. Let’s ask her.” So they ask the mom, and the mom says, “I don’t know either. That’s how grandma cooked fish.” So one dinner, they finally ask grandma. Her face changes. “You’re still doing that? I only cut the head off because we were poor, and the pan wasn’t big enough.”

That is most inherited rules. They were real solutions once. In old China, you drank warming teas because it was cold and falling sick could kill you. You didn’t bathe because you bathed in the river, and river water in an open wound was dangerous. Those grandmothers were not being foolish. They were solving the problems in front of them, with love.

But the problems are gone and the rules kept marching. We checked every one of those confinement rules with our doctors. The verdict on showering: not just allowed. Good for recovery. In today’s world, NOT showering for a month is the health hazard.

Most traditions are solutions to problems that no longer exist.

Which rule are you still obeying because the pan used to be small?

I promised you these letters would show the losses next to the wins. Here is the honest board.

The Loser: Back in #003 and #004, around Christmas, I told you I put 5% of my net worth into Midnight, the privacy blockchain. Since I went in, it’s down about 50%.

I’m not entirely surprised. Crypto swings like this. My thesis hasn’t changed: as institutions come on-chain, they need selective privacy to satisfy regulators, and I still believe privacy is a coming narrative. My timing was just bad. I’m holding.

The Winners: A couple of months ago I told you I took a position in Cloudflare (NET) around $220, betting on the wave of agentic AI traffic that has to pass through their pipes. Then the CEO announced a layoff and the stock dropped about 20%, which made no sense to me. The numbers were doing well and nothing about the thesis had changed. So I bought the fear: a second LEAP call at $195.

NET closed this week around $276, so the stock itself is up roughly 25% from where I entered. My positions are up a lot more than that, because I didn’t buy the stock. I bought LEAP calls: long-dated options, a leveraged bet that multiplies the move in both directions. That leverage is the whole point of the instrument, and it’s also the risk. The $220 call is up 54%. The $195 call, the one I opened while everyone was panicking, is up 127%. Both about three months old.

The Exit: UiPath (PATH) is now fully out of the portfolio. Regular readers saw me start that rotation in #020, so this closes the loop, and the real reason matters more than the trade: I could no longer explain to myself where its future growth comes from. I won’t stay invested in something I can’t explain. That dollar now works in things I actually understand.

The Confession: AI made building things so fast that shipping became a dopamine machine. An idea becomes a working thing in minutes. For a brain like mine, that’s not productivity. That’s a slot machine that pays out in prototypes.

The Cost: because AI lets me run five things at once, I did. And I let it take over more areas of my life than I want to admit. The research on multitasking is brutal: every switch between tasks burns focus on re-loading context, so you feel busy while doing everything a little worse. Splitting yourself across ten agents feels like superpowers. Your attention is still one lane wide.

The Fix: every day, I write down ONE thing. Not a to-do list. One line: today is a win if this one thing is done before the day ends. I can still run multiple projects around it, but that one thing is sacred, and everything else is bonus.

Nothing about this is new. It has always been focus. What’s new is the price of losing it. AI removed the bottleneck of doing, which means the bottleneck moved to attention. The scarce asset in your company is no longer hands. It’s your one lane of attention.

As promised, my private notes. No fluff.

See you next week,

Gabriel Judah

P.S. The twin, the AI CMO from last week, the agents that clean and publish my call replays overnight: they all come from the same playbook. Inside Two Hour CEO, over 200 members are building these “Loops” step by step, without being developers. Grab a seat here.