Every week there’s a new model, a new framework, a new agent library that promises to change everything. It’s easy to spend all your time keeping up. But if you focus on what changes all the time, you’ll always be chasing.
I try to start somewhere else: with what doesn’t change.
Principles first
What stays constant, in my experience, is you. Your experience, your judgment, your human take on a problem, and the way you do the work. Models come and go. The ability to understand a messy situation and decide what to do about it doesn’t expire with the next release.
The tool is not what makes the work valuable.
The tool is not what makes the work valuable. It never was. The same model in two different hands produces two very different results, and the difference is everything around the model.
Nobody buys the tool
At the end of the day, clients don’t care about the tools you use. If AI is the driving force behind a 10x improvement in some process, that’s great, but it isn’t what the business owner or the engineering team is paying for. They care about results.
Results come from good decisions: understanding the problem deeply, choosing the best approach, and picking the right tool for the job. Sometimes that tool is a large language model. Sometimes it’s a SQL query, a scheduled script or a better form.
When AI gets in the way
I’ll go a step further. AI is powerful, and for some teams it becomes an obstacle. The project starts from “how do we use AI here?” instead of “what’s actually wrong?”, and everything after that inherits the mistake. Just because you have a hammer doesn’t mean everything is a nail.
Learn the new tools. I do, every week. Just don’t mistake them for the work. Understand the problem, make good decisions, choose the right tool, and the results follow. That part hasn’t changed in fourteen years of doing this, and I doubt the next model will change it.