[7/24] 2026 Goal Progress

Another progress check on the goals I've been declaring. This week I was fresh off surgery, so I kept things easy on the body. Which meant my head was pretty much full of one thing the whole time: how do I get AI to do the work for me.

Health

Fresh off surgery, so the gym is on hold

I got the stitches from my hand surgery taken out. I've been told not to do anything high-impact for two or three weeks, so the gym is on hold.

Not exactly a substitute, but I did play a whole lot with the family. So in that sense, I figure I was moving around a decent amount.

I can also go to the chiropractor again now. I've been getting my neck sorted out, and I'm still keeping up the neck rehab.

Sleep, and an early-rising habit that's starting to slip

I signed up for U-NEXT recently. And then I started watching "Silicon Valley," a show I used to love. The result: I'm staying up a little too late.

The sleep itself is solid, uninterrupted hours, so I figure it's fine. But my early-rising habit is starting to slip a bit.

Work

"How do I get AI to work while I sleep"

This week too, I was basically getting work done with AI as usual.

On top of that, here's what I rolled out to the team again: how do we get AI to do, while we sleep, the work we were planning to do during the day. Let's refocus on that, is what I told them.

Let's aggressively kill off the situation where AI just sits idle unless a human hands it an instruction.

In my private side projects, I've actually got this system working pretty well. But at work, you can't exactly deploy willy-nilly. So the idea is: for the parts you can, set them up to run on their own overnight, then when you wake up in the morning, you check the results. You take responsibility for them. And you sharpen up the resolution of what you'll have it do the next night. Let's build that kind of cycle — that's the conversation we had.

AI is brilliant. But don't trust it too much

On the flip side, something else really hit home. AI is brilliant, but you can't trust it too much.

We run multiple projects with a small team. Partly because of that, we hit situations where the AI was wrong — and even though I'd noticed it was wrong — the work just carried on anyway. Honestly, we got right up to the edge of a real accident.

On the human side, we do the work of getting things down into a prompt harness. Even so, the shortcomings never stop coming. Catching that "gap of what isn't done" — that's still a human's position, for now.

The one making the final call is a human. That doesn't change. So I need to properly get back that feeling of AI just helping me out. ...In the context of work, that is. In other domains, "eh, just let AI do all of it" is totally fine by me.

Money

US stocks on standby, the bot at break-even

On the US stocks front, I'm waiting for the right timing right now.

The crypto auto-trading bot, honestly, is grim. June was great, but July's been bad, and it's gotten to the point where June's gains have basically been wiped out to zero.

But this is where full automation shines, I think. If I'd been doing this by hand and a month of gains I'd carefully stacked up got blown away, my mental state would probably take a hit. But since the bot's the one running it all on its own, it honestly doesn't hurt that much. If anything, it feels mentally healthy.

Two and a half months running it with real money, and the balance is break-even. But as a result, I actually think it's pretty good. The small position size might be a factor, sure — but in percentage terms, 0% over two and a half months isn't a bad number, is how I see it.

More than moving abroad, it's my English that's progressing

As for moving abroad, I'm not really thinking about it right now.

But I am keeping up the English study, properly. Duolingo is at 204 days straight.

And back to "Silicon Valley." Last time I watched it, I was leaning on the subtitles the whole way through. This time, it just comes in through my ears. I've even gotten to the point where I naturally think, "ah, so that's how they translated that." That's quietly satisfying.

Even watching a movie with my wife, I've started to really feel it — "hey, they just said this," and the fact that there are jokes only someone who understands English will get.

Other — maybe what I really want to do is "make things"

Thinking through a bunch of stuff this week, here's where I landed.

Getting AI to work while I sleep — doing that in both my private life and at work, while also having it help make life itself better. I feel like I need to think about that direction a lot more seriously.

Right now it's only software. But someday, I figure. Humanoid robots are getting a ton of buzz over in China and such, and robots that fold laundry and the like are being shown off, too. Something that takes over a bit of the housework and childcare. I want to widen my view to that kind of thing too.

That said, we've got a little kid at home. If a robot exploded or hurt them, that'd be scary, so for now I'm holding back and watching. Still, the parts I can do with software, I want to go at hard.

An aside. I've been using the 3D printer to make things my kid says they "want." What I realized is: on stuff like this, I pour in money without begrudging a single yen. And I think that's a really good thing. I want to see my kid's smile, I want to make someone happy. That side of me is winning out, is what it means, I think.

And it struck me. What I truly want to do — even among making things — sure, I love software and systems too, but maybe it's actually the offline kind of making, the kind you can touch with your hands.

Closing

Next week, I still can't hit the gym. So instead, I want to up the walking and stretching, and at least not let my habit of moving my body lapse.

At work and in private, the theme stays the same: "how do I get AI to work while I sleep." The parts I can do with software, I'll keep pushing on.

That's the kind of week I want to make it.