I’ve been tinkering with tools for as long as I can remember. Not in a "let me optimize my workflow" way — more like I can’t leave them alone. I was an early Evernote adopter back in 2008. I went through a whole Getting Things Done phase, cycling through apps and systems trying to find the one that would finally make everything click. I’ve tried more productivity tools than I’d care to admit.
There’s a saying I used to keep in my head — something like "a fool with a tool is still a fool." I can’t remember the exact wording anymore, but the idea stuck: the system doesn’t change the person using it. It can only do so much.
And I think that’s true. But I’ve always been more interested in the other side of it — the possibility that if you can figure out how to make a tool an extension of yourself, not just use it but actually integrate it into how you think and move through the world, how much more could you uncover? How much more could you unlock? That question is what keeps me tinkering. It has for years.
Two things in my life right now are testing it in very different ways. They don’t look alike at all, but they changed the same thing: what I pay attention to.
The wearables thing
A few years ago I got really into health tracking — Apple Watch, Oura Ring, Whoop. Before that, my approach to health was honestly pretty haphazard. I’d have stretches where I was on it and stretches where I wasn’t, and there wasn’t much connecting the two.
What changed was having data to look at. Graphs. Progress. Scores. I wanted a good sleep score, so I started going to bed at a reasonable hour. I wanted to keep my running streak going, so I kept running, and then I watched my fitness score climb. One thing came after the other — not because I had some grand plan, but because the feedback loop made small actions feel like they counted. A nap moved the needle. A walk moved the needle. A step in the right direction was literally and figuratively a step in the right direction.
I haven’t been able to get back to that peak, if I’m being honest. That was a few years ago, and what I did then apparently doesn’t work the same way now. That part is uneasy — trying to figure out what will work, and the fear of needing to come to terms with things just being different.
But what I took from that time is that I can’t just tug at one thing. It has to be holistic. So now I’m trying to layer things on slowly, one at a time, using different tools to capture information that gets me motivated. The tools didn’t give me discipline I didn’t have. They gave me visibility into what my small actions were actually doing, and that changed what I was willing to try.
The wearables shifted my focus from outcomes — am I healthy? — to inputs — what are my small daily actions doing? That’s the shift.
The agents thing
I used to prepare for big projects by studying. If I knew a project intersected with a particular area of software design, I’d spend the weekend reviewing those systems, re-reading a book on Python design patterns, thinking about how to architect things using best practices. That was how you got ready.
Now, when a big project is coming, I’m not watching design pattern videos. I’m looking at how people are setting up their systems so that AI coding agents can work more seamlessly — how to reduce slop, how to structure things so the agent produces something you’d actually want to keep. The agents shifted my focus from writing code to designing the conditions for something else to write it. Same shift as the wearables: the tool changed what preparation even means.
And the results are mixed. I’m probably getting more output overall, but it doesn’t feel efficient. There are so many more artifacts to parse through and throw out — markdown files, back-and-forth chat sessions, dead-end coding attempts. Sometimes you’re so deep into a session with an agent that it feels like it would take more energy to stop and go back to the traditional approach — which is just thinking for yourself — than to keep going.
I’ve been watching YouTube videos where experienced engineers are admitting they haven’t written a line of code in months. The things I was trying to do to reach that kind of status are completely different now. Now I need to learn how to be as efficient as they are at putting out quality code that isn’t written by me and isn’t full of slop.
The paradox of knowledge
Here’s the part that gets me, though: you still need to know those design patterns. You need to understand how systems should be built so you can train your agent to write code the way you would. The knowledge doesn’t become irrelevant — it shifts from something you apply directly to something you teach. You can still learn along the way; you can have the agent write up primers on concepts you want to use. There are ways to keep learning.
But learning has always been best done hands-on. And that’s the part you kind of lose. If you aren’t doing the struggle yourself, do you eventually lose the ability to teach the agent how to do it right? The wearables version of this would be: if the tracker tells you what to do and you follow it long enough, do you lose the instinct to listen to your own body? In both cases the tool is a proxy for something you used to do yourself. The question is whether the proxy replaces the skill or just relocates it.
What the good ones have in common
The tools that actually stick — the ones I keep coming back to, across all the years of tinkering — are the ones that change what I pay attention to. In both cases, the tool didn’t do the work for me. It moved my attention to a higher-leverage place.
A tool can’t give you something you’re not willing to build on. But the ones worth keeping aren’t just doing a task faster — they’re teaching you something about the task you didn’t know before you started.
That’s what keeps me tinkering.