What Miso Taught Claude Code's Creator
Making miso might look boring to the untrained eye. You cook soybeans, mix them with Kōji (fungus) and salt, pack the paste into a container, and then you wait. White miso takes a few months. Red miso takes 18 months to a few years.
I've lived in Japan for twelve years, so I've watched people make miso. The preparation is precise but unremarkable. Once it's packed, there's not much to do. It's a commitment to time, to a process you can't rush, to the idea that something worth having might take longer than your attention span wants it to.
These days, I keep bumping into this theme in an unexpected place: people building at the frontier of AI who spent time in Japan and came back thinking differently. Not in a vague "zen and the art of" way, but in specific, practical, I-changed-how-I-make-decisions ways.
---
Boris Cherny is the creator and head of Claude Code at Anthropic. Before joining, he was living in rural Nara with his wife, working remotely for Instagram at Meta. He was, by his own account, "the only engineer in the town" and "the only English speaker in the town."
On his blog, Cherny wrote that in his first two weeks in Japan, working from Nara with zero timezone overlap and almost no meetings, he "landed more code than I had in the previous year in Menlo Park."
Recently, on Lenny's Podcast, he described the life he'd built there: biking to the farmers market past rice paddies, trading pickles with neighbors, learning to make miso.
> Miso is this interesting thing where it teaches you to think on these long time scales. That's just very different than engineering, because a batch of white miso takes at least three months to make and a red miso is like 2, 3, 4 years. You just have to be very patient.
At Japan's seasonal pace, social life revolves around what's at the market this week, everything moves with the calendar rather than against it. He started reading sci-fi, thinking about long timelines, and eventually decided he needed to work on AI safety.
> I was living in this rural place. I was thinking these long time scales because everything is just so slow out there, at least compared to SF. All the things that you do are based around the seasons […]. You go to the farmers market and it's pimiento season and you know that because there's 20 pimiento vendors, and then the next week the season is done and it's grape season. I know how this thing [AI] can go and I just felt like I had to contribute to it going a little bit better, and that's actually why I ended up at Anthropic.
As if the miso, the quiet, the slower pace had collectively rewired his sense of urgency toward the deepest, hardest problems.
Qasar Younis is the co-founder and CEO of Applied Intuition, now valued at $15 billion. Before Harvard Business School, and his COO position at Y Combinator, he was an automotive engineer at Bosch, and spent time in Japan.
On his recent appearance on Lenny's Podcast, Younis explained why his team cleans its own office:
> Sometimes people will come to our office and they'll say, 'Oh, it's such a clean office, you guys must have like this giant cleaning staff.' And it's like, actually, we clean our office. Just like in Japanese school, as I mentioned, I lived in Japan, the students clean their own schools. We have a cleaning zen every week and everyone cleans the area around them.
The practice is called _souji_. In many Japanese schools, students clean classrooms and shared spaces themselves as part of daily school life. You take care of your own space.
> I think there's a direct line between: be quiet and alone, and clean your desk, and well-written software. And I don't know what that thing is, but it all falls in the same arc.
Maybe the connection between stillness and craft is so intuitive to anyone who's spent time in Japan, and so hard to articulate outside of it?
---
Llion Jones is a co-author of the 2017 "Attention Is All You Need" paper, the one that introduced the Transformer architecture behind every major AI model today. He's Welsh. He moved to Tokyo after taking a holiday there while working at Google, and when he quit to start his own company, he stayed in Japan.
Sakana AI, which he co-founded with David Ha and Ren Ito, is now valued at $2.65 billion and is one of Japan's fastest-rising AI startups. The name means "fish" in Japanese. The logo draws on the image of a school of fish forming a coherent whole from simple rules, echoing the company's interest in evolution and collective intelligence. The red fish breaking away signals a desire not to follow the crowd, but to pursue what comes next.
It's a way of thinking about AI rooted in collective intelligence and natural systems rather than brute-force scale, and it also fits a broader Japanese comfort with systems thinking, from manufacturing to ecology.
Jones has highlighted a mantra from engineer Brian Cheung: "You should only do the research that wouldn't happen if you weren't doing it." That sounds easier to do when you've put an ocean between yourself and everyone else's roadmap.
---
So maybe these are the questions for the rest of us: Is it about the ocean, or whether we just need the permission to stop swimming in the same direction? What would change about the way we build if we let things ferment longer?
Cherny's most productive coding stretch happened when he removed himself from the Bay Area entirely. Jones chose to explore what might come after the Transformer from Tokyo, not Mountain View. Younis built Applied Intuition hidden from the public eye for nearly a decade. How much of the AI industry's extreme geographic concentration is a strength, and how much of it is everyone drafting off the same assumptions?
Matt Alt recently explored how Japan's distance from the AI frontier lets people engage with the technology experimentally rather than defensively. Is there something in that distance, in being slightly behind the curve, that creates room for the kind of thinking these three are describing?
When Lenny asked Boris Cherny what he'd do after AGI, Cherny didn't hesitate: "I'd probably be making miso."
← Back to all posts