If you work in almost any white-collar industry today, keeping up with artificial intelligence has become a strange daily routine. You do not need to be a software engineer or a tech specialist to feel it. Every morning, standard office workers, analysts, managers, and creators open their laptops to face a continuous stream of updates, new model announcements, and endless video tutorials claiming that the tools they settled into six months ago are already outdated. Trying to

People claim they built an app with a single prompt. Usually that's an exaggeration. I've never built anything meaningful with one prompt — not an app, not a model, nothing that mattered. But I do write blog posts with a single prompt, and on a good day I finish fifteen of them in a day. The reason isn't the blog post. It's the groundwork sitting behind the prompt. Two things exist before I write anything: code that computes the numbers, and skills that turn a number into a

This dynamic reminded me of a beautiful essay by Bertrand Russell called "Useless Knowledge." In it, Russell talks about how sweet apricots were first brought to Europe from China. He explains how the word "apricot" has this fascinating, winding history: it was carried by Arab traders, derived from the Latin word praecoqua (meaning "early-ripe"), adapted into Byzantine Greek as praikokion, transformed in Arabic to al-barquq, absorbed into Spanish as albaricoque, and eventuall

When you ask an AI to help with a task — writing a document, building a spreadsheet, producing a report — it has no idea how you do things. It gives you something generic. The skills folder fixes that. It is a simple folder on your computer where you store written descriptions of how things should be done. The AI reads those descriptions before it starts working — not after, not when you remind it, but automatically every time — and the output matches your standards from the

In this article, I describe my experience of working with AI on various aspects of regulatory capital and how one formula triggered the "Pattern Recognition".

If you have been following my posts, you know that I am deeply immersed in writing about the Basel reforms. You might also have noticed that I use AI extensively in my calculations and code building. Today, I want to share my personal experience of diving into regulatory rules and understanding the underlying design concepts with the help of AI. This has become a normal routine for me—a way to ensure I don’t just memorize formulas, but truly grasp the logic and mechanics of t