[8/21] 2026 Goal Progress

The recurring post where I check in on the goals I've declared. This week was basically "back to work after the trip."

Health

Maybe the trip wore me out — I kept falling asleep early

This was my first week back after the trip. Maybe the fatigue was still hanging around, because I kept konking out pretty early.

At the very least, I never stayed up past midnight. My wake-up time didn't really change, but I was getting 8 or 9 hours of sleep, so my head was reasonably clear during the day.

Had a bit of a headache, though.

No gym yet. But my hand is healing up nicely

I still have hand surgery coming up, so I haven't been able to go to the gym.

That said, it's healing well. It's been over a month since the surgery and the swelling has gone down a lot. No pain in my hand, and putting load on my wrist doesn't hurt either.

So little by little — core training, lifting light dumbbells, that sort of thing. Next week, I think.

Also, on the trip I was mostly snorkeling, and swimming really does tighten you up. There's a facility near me with weights and a pool right next door, so I'm thinking about making a habit of going there regularly.

Quitting smoking: I bought a time-lock box and cut my daily count by 50%

The smoking progress is the fun part this week. I started something the day before yesterday.

I was going to build a time-lock box myself. But there was the peripheral hardware, the soldering, and the risk of the 3D printer filament melting — so I decided to just buy a finished one for now.

I actually bought it, and I've been doing this thing where after I smoke, I lock it away for 30 minutes, then another 30 minutes, then another hour. Doing that, my daily cigarette count dropped by half — a 50% reduction. I think that's really good.

Outside of work I don't smoke much anyway — literally just a few. What I wanted to fix was the situation where I smoke way too much during work. So cutting that by 50% is huge.

An IQOS pack runs me about 600 yen, so when I think about 600 yen a day going out the door, that's pretty significant.

Work

I got serious about "what actually needs to happen today" — and finished in half a day

In the blog post I wrote two weeks ago about "working in the AI era," there was this point that humans are the ones generating the human bottleneck. That felt backwards to me, so I got serious about properly identifying what today's tasks actually are.

And it turns out — it really does finish in half a day.

The leftover time went into things I like, research that could contribute to the company, and the kind of organizational-reform conversations and long-term "where is this company going" thinking that you'd never have the headspace for if you were genuinely swamped. I got time to actually think about that stuff.

None of it was screaming-high priority, so when interrupt requests came in over Slack, I'd just toss them at the AI and keep things moving. And then actually verify the behavior properly. Because my plate wasn't overloaded, I could move safely, and honestly I was in a much more stable headspace too. It was really good.

Put a schedule down, even a provisional one

I've already been rolling this out to the team.

Whatever it takes, put a schedule down — even a tentative one. To figure out what needs doing today, you need some form of schedule to exist. Then, against what you've put down, you think through how to actually make it happen — the same way you always would.

Do that and "oh crap, I have to do this today" becomes unmistakably visible. If it isn't visible to you, I think you're being fairly passive.

So people who move proactively — they end up looking busy because they keep taking on work themselves, but that posture is exactly what makes life easier for them. That's the kind of world we're in now with AI, I found myself thinking.

Turning meeting notes into a provisional schedule (idea stage)

Haven't gotten to automation yet.

The "figure out what needs doing today" thing has spread through the company and is getting genuine buy-in. And to make the provisional schedules — and the fixed ones — actually happen without dropping anything, you need a mechanism that says "here, do this."

The idea that came up: have the AI take notes on meetings where people are actually talking, automatically identify the action items, build a provisional schedule, and then ask a human PM to review it.

The catch is that having a client's meeting recorded by an AI without asking is a non-starter at some organizations.

So there, you keep doing it the old way — a human quietly takes light notes in the background. And for decisions made during the meeting, it gets a bit multitask-y, but you post the announcements in Slack behind the scenes: "we got the go-ahead on this," "please start prepping," that kind of thing.

Building a mechanism that lets you do all that smoothly while participating in the meeting feels pretty necessary. This was a week where a lot of ideas like that came up.

My head is stuck on organizational theory

The actual work really does get done by the AI now, and my job is evaluating that output and taking responsibility for shipping it. That's genuinely becoming the main job. And I'm fine with that — that's where a human should be the bottleneck.

Which means — I mean, the AI implements fast, and me reviewing it afterward, pointing things out, and having it fixed is also fast. So my attention has shifted more toward time spent with people than it used to be.

Communication with people, someone's attitude, what they said — I've gotten more sensitive to all of that than before. And it's made me feel I need to think again about what a healthy organization actually looks like.

Not that I'm struggling, though. We've got great people.

There's some communication loss and dropped balls here and there — everyone's human. But communication that costs you for no good reason is something we really should be fixing wherever possible. That's the organizational-theory headspace I'm in.

Money

Migrated a good chunk from PayPay Securities to SBI Securities

I got a lot migrated from PayPay Securities to SBI Securities.

The NISA account is still on the PayPay side, though, so the monthly contributions are still running there. When the timing's right, I'll look into the proper procedure — whether this year's remaining four months' worth can be carried over, that kind of thing — and ultimately consolidate the brokerage side onto SBI.

And over there, US stocks. I've got my own hypothesis about which sectors will still exist in 10 years, so I researched companies that look likely to shine in those areas together with the AI, picked out some targets, and actually went through with buying a few shares.

Now I just wait for the price to go positive. Small steps, sure, but I'm getting it done.

The trading bot: great in June, terrible in July. At least I've got the analysis

On the crypto auto-trading bot: June was excellent, July was awful.

August was also completely dead right up until this week — and then from the back half of this week it suddenly picked up.

Roughly, the character of the market is that the bot rakes in profit when volatility is high. And July was pretty stable, honestly. So it couldn't detect the trend reversals and ended up orphaned.

Ten steps forward, eleven steps back — slowly bleeding into the red over time. That's the analysis I managed to put together over these three months, so if there's a countermeasure I can implement, I'll implement it.

The result is negative, but it's not a meaningful amount. Not at a price point that wrecks me mentally. I said I'd do this for three years, and I don't think I'm even 10% of the way in, so I'm going to keep at it.

Saving money: 50% less tobacco, and quitting UberEats

On the savings side — it's only been two, three days — but tobacco is currently on a cycle that saves me 50%. That's pretty big. I get health and money at the same time.

To a non-smoker this probably sounds like "what on earth are you talking about," but being a smoker is actually kind of rough.

The other thing: I eat out a lot, I use UberEats a lot, and I've got a strong feeling I'm throwing away money on it.

Since I've got work reasonably under control now, I want to start making my own lunch, cooking for myself — and put everything I save straight into investments.

For the longest time it felt like I had no breathing room, when actually I did — I was just leaning on UberEats because I found it a hassle. So from now on, UberEats is the last resort.

That said, convenience-store food loaded with additives isn't great either, so yeah — cooking for myself it is.

English & Moving Abroad

232 days straight on Duolingo. And still nothing comes out reflexively

Duolingo is at 232 consecutive days.

But like I wrote in the last post — in actual situations the words just don't come out, or they come to me after the fact. There's still a version of me whose English doesn't fire off reflexively.

So I'm building an English learning tool myself

I've got an idea, and I'm implementing it right now: I'm building my own English learning tool. I'll use it myself while building it, and if it turns into a business, lucky me.

The core problem with how Japanese people learn English is that thinking in Japanese first and then speaking English is the biggest bottleneck. And in my own case, I did all the grammar and vocabulary cramming, but I can't use any of it.

So what's needed is the ability to combine things and speak on the spot, and I want to build a tool that enables that. None of the details are implemented yet, so I'll figure it out as I build.

Also, in building this application, I've been making a ton of Claude Code plugins and skills and using them as I go. I'd like to write about that on the blog at some point.

Other

Hanzawa Naoki turned out to be the bible for working in the AI era

I don't really know why, but lately I've been going along with whatever pops into my head.

And out of nowhere, Hanzawa Naoki — the Sunday night drama, there were two seasons of it — just popped into my head. So I hunted down somewhere I could watch it and rewatched it after all this time.

And yeah — as a study in how to work in the AI era, Hanzawa Naoki is honestly kind of a bible.

The extreme outbursts and the raised voices, I can't really get behind. But how he deals with people, the returns that come from that, the importance of colleagues, trust, integrity.

And since it's set in a bank, there's a lot of language, honorifics, and etiquette that comes with hierarchy. I think the kids entering the workforce now would do well to watch it once.

People follow the person with room to spare

I mean, it's a drama, so I'm sure reality is nothing like it.

But Hanzawa is basically getting set up by everyone, constantly. So he's having an incredibly hard time — and yet he throws himself fully into his colleagues' problems, keeps worrying about them, and still doesn't let his own work slide. And he doesn't blow up in a rage while barely hanging on.

He's got so much room to spare, and such a big heart.

That posture is what makes people follow you — I got shown that pretty objectively. And in my own working life so far, I've definitely felt that those are the people everyone gravitates toward.

Am I that person right now? ...Yeah, I can't say yes.

So that's something I want to be deliberate about, using him as a reference.

Wrapping Up

For next week: I had two, maybe three new ideas come up this week, so I want to take a bite out of those.

Also, starting next week I'm going to restart exercise and strength training. Home workouts first, and after that I'm considering going to the sports center to swim, that sort of thing.

Then, keep the English learning streak going. And keep the time-lock box quit-smoking thing going, no matter what.

That's it.