[9/25] 2026 Goal Progress

The usual round of writing up progress on the goals I declared. This week was Silver Week, and I basically spent it moving around nonstop.

Health

Silver Week, and I was genuinely active

It was Silver Week, so the default plan was going somewhere with the family.

It was right after a typhoon, but we went out to Chiba to hang with a friend's family, and hit a horror festival that was running in Ueno. We also spent a whole day at Legoland.

So yeah, I was pretty active — walked a ton, got a real amount of exercise in.

My health has improved a lot too. I'd been worried about whether I'd get through the break in one piece, but it's down to a runny nose and a slight cough now. It's cleared up.

For some reason I wake up earlier on the couch

Sleep was decent too — in bed by 23:00 or midnight, up at 7 or 8. That worked out well.

Here's the thing though: I can't get up when I sleep in the bed, but on the couch I actually wake up early. And I don't feel sleep-deprived. My body hurts a bit though.

So lately I've been thinking maybe the couch is just better for me.

Quitting smoking: the time-lock box habit is completely gone

Quitting is a total failure right now.

The time-lock box is down to maybe one or two uses a day, locking it for an hour at most.

The old rhythm — smoke two, lock it an hour, open it, smoke two, lock it another hour — that habit is completely gone.

Work

The long weekend cleared my head, and the path got clearer

I barely worked this week, so there isn't much of a highlight.

My head had been pretty cluttered lately, and the long weekend clearing it out got my thinking organized.

Which meant that, factoring in the AI era and the specific character of our company, the path for which concepts to think about, discuss, and solve problems around got a lot clearer. That was the big one.

Efficiency matters as a concept, sure — but I'm coming around to the idea that efficiency followed by delegation is the real key point.

AI employees are now usable day-to-day

On the AI employees that minimize human bottlenecks as much as possible, there's real progress.

I've been hammering on them locally and they've gotten to where I use them day-to-day.

The bare minimum is working, so once I polish them a bit more, the next step is rolling them out internally.

The problem: an organization that can scale out AND scale in

No matter how well the AI, the workflows, the systems all run — we're not a company built around focusing on its own product. Our workload has enormous swings.

Given that, being an organization that can scale out and scale back in has become something of a defining challenge.

The simplest way to scale out is to just hire a bunch of people. But if they're full-time employees, you can't scale back in. Not easily.

And even with contractors — once they've helped you through a scale-out, you end up thinking "they really saved us," "I want them to stay." So scaling in gets hard.

You could just stop giving them tasks, but then when they've drifted away and you genuinely need them again, that gets rough.

So the concept of employment itself is something I need to rethink.

AI can't take responsibility. So the bottleneck concentrates on people

What does it actually look like to scale out and scale in? I suspect the part that expands and contracts has to be AI.

But doing it perfectly, with zero waste, is impossible. Why? Because AI can't take responsibility.

So the bottleneck concentrates on the people who do take responsibility. It already does, right now.

And as volume grows, as projects multiply, that scales up proportionally.

So to distribute it, you hire people who combine a high level, high skill, high communication ability, high judgment — and you make distributing evaluation and responsibility the point of your hiring.

Money

Took profit on my biggest US stock bet. It covered rent

On US stocks — the one I'd bet biggest on went way up, so I took profit.

It came out to roughly covering rent, which felt great.

On the target I set this year for pre-tax gains from stock trading, I'm at about 95%. With three months left, it's less "hit the target" and more "it'd be nice to get there just by playing it normally."

The crypto bot: experienced every pattern, still negative

On the crypto bot — I'm fairly sure that over these four and a half months I've hit basically every pattern.

Sharp rallies, sharp drops, crashes, spikes, range-bound markets, patterns with violent trend reversals — all of it.

Ran through all of it and came out negative, in a three steps forward, five steps back sort of way. So I've stopped it for now and I'm rebuilding the bot from zero.

Though when I say negative — it's about ¥10k down per month, so I'm out around ¥40k right now. Losing slowly and steadily.

Basically I've become a guy who just pays fees to the exchange, faithfully.

I'm taking it as a good opportunity, and I'm building the bot from scratch again.

Looking into tax strategy, but there aren't many options

Not much on cutting costs. I've been looking into tax strategy lately.

There really aren't many options. And buying a needlessly expensive house just for the tax treatment is idiotic. So I'm still looking for a smarter approach that's within the law and not in the gray zone.

On income: stocks did well this month. The bot is half a hobby, so it's negative, but whatever.

Otherwise it's just my regular salary from my main job. Side gig inquiries welcome.

Duolingo at 267 days

Duolingo is at 267 days.

But I'm only doing about 5 minutes a day, so progress isn't great.

As for moving abroad — not really an option right now.

Other

Just take action

The thing that stuck with me: taking a ton of action really does matter. Personal life and work both. No plan required — just find something on the spot and go on a spontaneous trip.

It's effort, but once you go, it becomes a good memory and you're glad you went.

You also learn what things cost and what's reasonable, it makes you want to keep taking trips like that, and it makes you want to work harder.

And unless you're gambling, it's hard to imagine taking a big loss from a big action like that.

To make it low-risk high-return, it's about acting small and whether you can look back and think that was worth it.

Just act — the odds of the result being genuinely terrible are basically nil, I think.

So: just take action.

Wrapping up

Next week — my health is coming back, so I want to actually do the at-home strength training properly.

Then, once I've shaken off the stiffness, restart the gym.

On work, I want to get the AI employees all the way to an internal rollout.

And there's jev going around — I want to do some research and evaluate adopting it, among other things.