I've been tinkering with AI lately, but not in the way most people approach it. Instead of thinking about how to make my existing workflows 10% faster, I've been asking myself: what mental models could get unlocked if I threw away what I know?
And it's a lot of fun.
Questioning Everything
Take pull requests. We've internalized them as the way to collaborate on code. But do we actually need them? What if we could run CI locally with a system that cryptographically attests that the checks ran against the exact code being pushed? You'd skip the entire CI queue, the waiting, the context switching. The attestation proves the work was done-why repeat it on a server?
This isn't about optimizing PRs. It's about questioning whether they should exist at all in certain contexts.
Anything is possible, really. Once you let go of "this is how it's done," you start seeing possibilities everywhere.
Follow the Builders
I struggle to process why people say they don't find AI useful. At the same time, I see gurus emerging everywhere-jumping onto courses, offering advice, building personal brands on the AI wave. It's common. I ignore them.
I follow the builders instead.
People who look at the future open-minded, willing to shape something better rather than just profit from the hype or stay relevant. It's a lot of fun watching those people work. Tobi from Shopify posted the other day how he built a tool to process some medical imaging format. Not because it was his job, but because he could. That's the energy I'm drawn to.
The Investment Mindset
I'm spending money on AI. A lot of money on tokens. I finally understand what Peter Steinberger meant when he said this is an investment. It truly is.
But we developers hate paying for things. I used to be like that too. We'll spend hours building something ourselves to avoid a $10/month subscription. We'll fight for open source alternatives even when they cost us more time than the commercial option ever would.
With AI, I've had to rewire that instinct. Every token I spend is buying me time, learning, and capabilities I didn't have before. It's not an expense-it's compound interest on my skills.
An Invitation
If you're still approaching AI incrementally-asking "how can this help me write code faster?"-I'd invite you to try something different. Ask instead: what assumptions am I making that might not need to exist?
Challenge the mental blocks. Question the workflows you've inherited. Look at things with an open mind, willing to throw away what you know if something better emerges.
The people building the future aren't optimizing the past. They're imagining what could be.
Join them.