Sep 9, 2026 · By Kiyo Darner · 5 min read
AI Is Not a Genie. The Hard Part Was Never the Data.
Last week I wrote about why I built shochumixer.com. This is the part I left out, which is what it actually took.
People keep asking how long the site took, and I think they expect me to say the website was the hard part. It was not. Building a website in 2026 is easier than it has ever been. The data is expensive and it takes time, but expensive and slow is not the same as difficult. You can solve expensive and slow by grinding.
What you cannot solve by grinding is the part nobody sees.
AI Cannot Make Something That Does Not Exist
I used AI to scrape. I am not going to pretend otherwise. But AI is not a genie. It cannot magically produce something that was never there in the first place. Ask it what to mix with a heavy imo shochu and you will get an answer that sounds right, because sounding right is what it is built to do. Whether the answer is any good is a separate question, and that question is a person's job.
Somebody has to decide the logic. Somebody has to work out that single distillation leaves the raw material in the glass, that a clean rice takes almost anything beside it, that a heavy imo will fight a mixer picked by analogy to vodka, that the koji changes the acid, that strength changes what survives dilution, that time in wood moves all of it again. None of that arrives on its own. I wrote it.
I spent hours on hours trying to think of as many variables as I could before there was a single bottle in the system. The site holds 543 bottles, 163 mixers and 170 distilleries, and the rules have to hold an opinion about how any of it meets any of the rest. Getting the records in was the easy half.
I Am the Weak Point
Here is the part I would rather not write. I am human. I make errors. I have biases. My palate is not perfect and it never will be. I forget things.
Every one of those is now baked into a tool that other people are going to use. That is uncomfortable, and it is also just true, and I would rather say it out loud than have somebody find it. If a recommendation on that site is wrong, it is wrong because I got it wrong.
I Also Have a Restaurant
This is a passion project with no backing behind it. I run Juraku full time, and running a restaurant is a full time job with its own list of problems that do not wait.
The way ideas work for me is that when one turns up I have to start on it immediately or the line of thought is gone. That is not always possible with a dinner service in front of me. A lot of this got built in the gaps.
So the site is not finished, and I am not going to pretend it is. It is ongoing. I will keep working on it and it will keep getting better.
Which Is Where You Come In
Share it if you think it is useful. And tell me where it is wrong.
I mean the criticism part. If you know shochu better than I do, and plenty of people do, I want to hear it. If a pairing on there is nonsense, say so. If I have a producer wrong, say so. The only way this becomes a genuinely good tool is people telling me where it falls down.
Why I Bothered
My goal has always been to share Japanese food and culture without a barrier.
There is usually a barrier. It sounds like "oh, you don't know how to use chopsticks?" It looks like people not understanding that Asian cultures are not one culture. And I will admit that when somebody bows at me with their hands pressed together, I am slightly offended, and somewhere in my head I am thinking what do you think I am, a monk.
But I hold it in, and I hold it in on purpose. That gesture is not coming from a bad place. It is coming from not knowing. If I snap at somebody because I think they should have known better, they will never want to learn anything from me again. Knowledge should not be gatekept. Not mine, anyway.
I just want people to enjoy my cooking. Everything else is extra.
That is the same thinking behind the site. Shochu gets sold as something you have to understand before you are allowed to like it. Learn the serve, learn the ritual, learn the words, and then you may have some. I think that is backwards, and I think it is a large part of why the category has not moved here. So the site starts from what you already drink and works back to the bottle, instead of asking you to earn it first.
Try it, and tell me what is wrong with it.
Keep Reading
Hungry? Come see us.
