One piece a week, released in order from the first issue. Subscribe and we’ll email you each new issue the moment it opens.
One report outside my field, and that single week was where it began. AI extracts; whatever must not be wrong, people verify to the end — the discipline that became the backbone of _y Tower.
Read the issue →A lab where real engineering keeps getting pushed back by a flood of requirements. How a team of three doing the work of ten rethought AI’s place — not as a tool, but as hands and feet.
Read the issue →Thinking we were short on people, I gathered as many as thirteen — but they scattered, and I was the only one left. Six months of filling that space with AI instead of people.
Read the issue →A story that began with one report led to the lectern and a book. A record of the one conviction I found at the end of it.
Read the issue →A two-month first-pass patent review, done in a day. Instead of asking AI for the answer, we had it erase what was ‘clearly not a match’ — and kept the risky 5% in human hands.
Read the issue →My job is reviewing other people’s beginnings, yet I kept putting off my own ‘someday.’ Why I chose a newsletter as the first step back toward the people I care about.
Read the issue →Finding certification requirements once meant digging through dozens of PDFs by hand. Now it takes seconds — and leaving AI out made it cheaper and more accurate.
Read the issue →From industry trends to the master schedule, field pilots, and an issue tracker — everything the lab runs, on one dashboard. What I learned building it, and what didn’t work.
Read the issue →All I built was one alert service for Korea’s public procurement portal — then data collection, payments, email, and marketing followed one after another. A record of eleven months of trial and error, and what came of it.
Read the issue →We couldn’t simply trust the list the patent team had filtered for us, so I pulled the data myself and built a database. When the review was done, ten new patents were left — and so was one question.
Read the issue →AI and I put about 30 apps on the app stores. Total downloads: about 3,000. The apps I was surest about were the ones nobody used.
Read the issue →Footballers train even on days with no match. Asked what founders train, I had no answer. Looking into the training YC and Stanford give founders, I came to see our own group differently.
Read the issue →I tried to hand everything to a single assistant, and it fell apart. Then I had my specialists hold meetings, and I got lost again. Only after falling twice did the frame of _y Tower remain.
Read the issue →Even with the org chart drawn, the results were uneven. To build an organization that fixes itself, I first needed a seat that doesn’t trust itself.
Read the issue →It wasn't a sophisticated intrusion. It was a door I'd left open and forgotten. And when I locked it, it was nearly my own connection that got cut off first.
Up next · Coming soon