19. Choose Boring Tech

Boring tech is underrated, because boring tech is free time.

Now, I’m the guy who tells everyone to always invest in the latest technology, and I mean it. So let me draw the line I actually live by. Use the newest thing where it’s genuine leverage, the part that multiplies your output. Use the most boring, proven, forgettable thing everywhere else. AI in my workflow is worth being on the bleeding edge for, because it multiplies what I can make. The framework my site runs on is not. One is my edge. The other just has to work and never wake me up at midnight.

I learned this the expensive way. I spent a chunk of 2024 trying to build my own AI agent framework, Pynions, at a layer of the stack where my competition was companies spending billions. It was a great way to learn and a terrible way to ship. The lesson stuck: don’t build the boring infrastructure, rent it, and put your energy on top, where the customer actually is.

Your users have never once cared whether your stack is cool. They care whether the thing works, whether it’ll still be there tomorrow, and whether it solves their problem. A boring, stable, well-documented tool gives you fewer dependencies, fewer moving parts, fewer late-night fixes, and fewer rewrites. Every hour you don’t spend fighting your tools is an hour you spend on the offer, the copy, the customer, the distribution, the parts that actually decide whether this works.

Choosing tech as identity is a trap. The cool stack doesn’t pay rent. The boring one you understand completely, that you can fix half asleep, that has ten years of answers already written on the internet, is the one that lets you move fast for years instead of looking fast for a month.

Pick the simplest thing that’ll still be maintainable a year from now. Make it boringly reliable. Ship on it. If you outgrow it, that’s a good problem, and you’ll have the customers to justify solving it.

What cool tool are you maintaining right now that’s adding complexity without adding a single customer? That’s the one to cut.