Building personal utility in the age of AI

One thing I’ve increasingly appreciated about AI-assisted development is how practical it can be for building small software. The world doesn’t need another social media app or crypto-tracker. What the world needs is more actual utility. Not more products or services. Tools.

This is not to say that products and services are bad, far from it. There just aren’t enough personal tools, or tools that can be made personal. Tools that can solve immediate problems and then get out of the way. That side of AI-assisted development gets overlooked. Too many people immediately frame it as a business opportunity instead of a utility. The unfortunate side effect is that people stop seeing small ideas as worth building. Not everything needs to scale.

I’ve been building a handful of tiny local-first utilities for the last several months. Mostly workflow stuff: cleaning pasted text, comparing text differences, organizing snippets, temporary notes. Small things that are immediately useful while working.

What I like about these tools is that many of them don’t really need much infrastructure. No accounts. No subscriptions. No syncing requirements. In some cases, no backend at all. Just a browser tab and local storage.

That kind of software feels strangely rare now.

A lot of modern software feels obligated to become a platform. Even very small utilities often arrive bundled with dashboards, onboarding flows, notifications, analytics, and recurring pricing. Sometimes that makes sense. Often it doesn’t.

I’ve found myself moving in the opposite direction instead. Smaller scope, narrower purpose, less noise. Simplify, simplify.

AI has made that process much easier. Not because it magically builds perfect applications, but because it lowers the friction between “this would be useful” and “this now exists.” AI-assisted development dramatically lowers the barrier to building small software. Too often those ideas become startup pitches or SaaS concepts with the thinking of “what do people need/want” instead of “what do I need/want.”

When I’m sitting at my machine and saying to myself, “sure would be nice to be able to ____ right now,” I can. I can build it quickly. Sometimes immediately. AI-assisted development makes personal utility much more achievable.

That utility changes the kind of software people are willing to make. Not just in terms of using AI, but in not linking databases, not building backend security, no updating API keys. No servers, no accounts, no credit cards. You can build a personal, useful tool that suits your own purposes.

You can build tools purely because they improve your own workflow a little. You can experiment with odd ideas that would never justify traditional development time. You can make software that serves a specific personal need instead of trying to appeal to everyone.

I think there’s real value in that direction, especially when paired with privacy-first or local-first design. Software does not always need to become a service. Sometimes it can just remain a tool.

Not everyone wants to start a software company. Some people just want better tailored tools. And honestly, I’d like to see more of that.