What AI Should Actually Be For: Software That Fits YOU

I have a friend who works in the restaurant industry as a prep cook. He's excellent at his job, and over time he figured out the most efficient way to do it. He had a personal workflow that played to the way his brain works, and it made him faster and happier. When the owner of the restaurant discovered his process, they got angry. They demanded he work the "correct" way, which was slower and made my friend actively miserable. The food still got prepped. The job still got done. But the end result was worse across the board, for everyone involved, because someone decided that uniformity mattered more than outcome.

I keep thinking about that story because of something Ron Cass said to me in a conversation last week. Offhandedly, he said: "Software should be flexible. It should fit the customer."

I think that captures what excites me about AI more than it terrifies me.

Here's the thing about human beings: we are endlessly, beautifully creative problem solvers. Give five people the same problem and you will get five completely different solutions, each one tailored to the way that specific person's brain processes information. If you've never watched Taskmaster (and you should, immediately, it is an incredible show, go watch it right now), the entire premise is built on this. Five comedians get the same silly task, and every single one of them approaches it differently. Some of them are methodical, some are incredibly chaotic, and some make it their mission to find as many loopholes in the tasks as possible. The best part is that most of the approaches work, even the unhinged ones, because people are wired to identify a problem and build their own path to a solution.

So why is software so bad at honoring that?

Think about the tools you use every day. How many of them force you into someone else's workflow? How many of them make you adjust YOUR brain to fit THEIR process, instead of the other way around? It's the prep cook problem at scale. The job gets done, sure. But it gets done ish. You're working around the tool instead of with it, and the gap between "ish" and "what I actually need" is where problem solving joy goes to die.

And we just... accept ish. We pay subscription fees for ish. We pay multiple subscription fees for multiple flavors of ish, from multiple companies, none of which talk to each other, all of which were designed for a generalized persona who doesn't exist.

That's the part that AI, applied thoughtfully, has the potential to break open.

I want to focus on “applied thoughtfully” here for a sec. AI applied lazily gives us the same rigid, one-size-fits-all garbage we already have, just generated faster. Lazy AI is what my dermatologist implemented, in which I still navigate a phone tree except I can’t interrupt it and it actively makes my call longer and more annoying. Lazy AI is Meta incorporating AI summarizing when my best friend sends me 10 excited Instagram messages in succession.

But AI applied thoughtfully? That puts the power to build back in the hands of the person with the problem.

I've been experiencing this myself over the last few months. I started noticing small annoyances in my daily life, things that bugged me just enough to wish someone had built a solution for them, and then I realized: I could build the solution. Not by going to find some "bespoke" app that costs $8/month and does 40% of what I need, but by sitting down and describing what I actually wanted. I built a personal recipe app because every recipe site on the internet is a nightmare of pop-ups and life stories before you get to the ingredients. I built a better PDF downloader for Figma because the existing one wasn't cutting it. I wrote Arduino code to turn my bed remote into a phone app because I kept losing the damn remote in my blankets and I wasn’t about to pay a subscription fee for an app for my BED.

Each one of those was a specific problem with a specific solution built for exactly one person: me. I don’t get prompted by these solutions to see their new features or get advertised to. There’s no feature bloat, no monthly fee (well, minus Claude Code, technically, but I’ll take that over 20 other subscriptions any day), no writing feedback emails to a product team begging them to implement a feature that I actually do very much need.

And look, I know the fear side of this too. I work in tech. I was laid off in February. I see the conversations about jobs and displacement and the worry that AI is coming for creative work. Those concerns are valid and they deserve serious conversation. But when I zoom out and think about what happens when enough people, both inside tech and outside of it, start building their own tools for their own problems? When a prep cook can describe the workflow that makes him fastest and happiest, and the software just... does that? I think we step into something that looks a lot less like automation and a lot more like a creative expansion we haven't seen before.

My friend is still at that restaurant, by the way. Still doing it the owner's way. Still a little less efficient and a little more annoyed about it every day, and the job gets done. But I keep thinking about what it would look like if the tools and processes met him where he was, instead of the other way around.

Software should fit the customer. We finally have the technology to make that true.