AI creates new problems. That can be solved with more AI.

AI is now part of my daily routine — be it to structure my work or to write code. And while it has brought my output to another level, it comes with its own set of problems: unparalleled programming speed leaves me with volumes of code I can barely peek into, random unpredictable errors hit where I least expect, and an agreeable interlocutor sometimes leads me straight down a silly rabbit hole.

Luckily, I have a very powerful tool to counteract all these shortcomings. I use AI to check it did not mislead me. It tells me all I need to know. All? Almost. At least, as soon as I master the ins and outs of context engineering. Then I’ll dig deeper into agentic review. And… well, the list is long. The details of each technique have had the best engineers busy for years.

Sometimes, I feel dizzy. I suspect I’m not alone.

Now I can confidently say that the quality of my work has improved drastically over the last months — and it keeps improving. Why? Because I don’t measure the quality of my work with AI. I know when a user is happy using a product that I shipped, and I know when a platform architecture is robust. There’s still a human in the loop — me. I still run that boat.