Those vacations were the most intense I’ve had in a while. Time off is always the chance to focus on building knowledge or skills, whether for personal enjoyment or for work. And this summer break was the perfect moment to disconnect from the hectic daily routine, the endless flow of emails and back-to-back meetings, and dive into introspection.
Walk the Talk
Our job is to look for AI-native companies (post-LLM startups), because we believe they are a different breed, born in a different world. Which begs the question: what about us? Why should funds be any less affected by this new paradigm? Could we be an AI-native fund, making moves as fast as the companies we invest in? And if we could start from scratch, where would we start?
Hongdae in Seoul, known for K-pop culture, nightlife, trendy shopping—and now, vibe-coding ✨.
August 2nd. I was alone in a tiny, windowless room in Hongdae, Seoul, while it was 35ºC outside: zero distractions, almost no family duty, and surrounded by konbinis open 24/7. It was the perfect setup to switch into full focus mode (some would call it *berserk* mode), locked in for almost 22 hours a day with bare sleep. I started testing nearly every tool, API, and flavor of AI coding - tap coding, vibe coding, agile coding, spec-driven coding, you name it - until I found my own pace and recipe. Then I literally went on a token frenzy, on a quest to turn our management company into an experimental sandbox with one clear goal: to make our workflow more efficient, more personalized, more agile, and more collaborative.
Thank god, nobody can see my spaghetti code 🙈
Tale From the Neolithic (circa 2022)
Sounds familiar, doesn’t it? It’s basically every company’s dream. And of course, it was our initial vision too, until reality struck. Four years ago, we started with a fancy yet incomplete home-made platform using Ruby on Rails, VueJS 2, and Postgres. I built it during the fundraising phase (yes, investors raise money too), but things went south pretty quickly once we began investing which left me absolutely no time to refine and upgrade the platform. A small fund faces the same workload and regulation as a large one, only with far fewer people to handle them.
I then decided to move to Airtable, which allowed me to add features quickly through scripting and required almost no maintenance. Our data feeding was inconsistent, and we believed a robust CRM was the answer. So we migrated to Pipedrive, with its out-of-the-box Google Suite integration, thinking this would finally bring auto-feeding.
Airtable is a great tool, but even if you script it to death, it remains very rigid at its core.
... continue reading