The Rise of Small Software

In The Rise of Personal Apps, I wrote about people building software for a market of one. There’s a related idea I’ve been living through lately: small software.

Small software isn’t an app you ship to the App Store. It isn’t a SaaS product with a roadmap, a login screen, and a growth team. It’s AI built software for solving a specific problem you or a small group of people who share that problem have.

The problems that never made the list

Every day I run into little frictions that happen over and over again on a weekly or monthly basis. A spreadsheet I keep reformatting by hand. A workflow between two tools that don’t talk to each other. A tracker that no commercial product quite gets right. A shared system for coordinating something with friends that always ends up as a messy group chat.

Historically, those problems sat on a mental backlog that never got touched. Not because they were hard in principle, but because the ROI for automating them was terrible. Building even a rough tool meant evenings of setup, debugging, and half finished projects. By the time I got something working, the original annoyance often felt smaller than the cost of automating it.

AI changes that ROI. I can now tell ai to build something while I work on other things, so these small automations get built in the background. The constraint wasn’t talent or ideas, it was time. This will lead to a rise in small software.

Not for many people, but for a precise few

Commercial software has to solve problems that are common enough to support a business. That means averages: features for the median user, integrations for the biggest use cases, polish for support tickets and scale.

Small software flips the target. It doesn’t need to serve many people. It needs to serve you, or you and a handful of others, in the exact shape your lives already take. Your naming conventions. Your calendar quirks. Your weird combination of tools or workflows that are unique to your daily/weekly/monthly life.

AI is unusually good at building these prototypes in a background. You describe the workflow in plain language, iterate on the output, and reshape it until it fits. On top of that, the software doesn’t need to be written so it scales to millions of concurrent users. It just needs to work.

Why the time ROI finally works

Small software doesn’t need to be perfectly coded to scale. It needs to be functional and accessible to the people who will use it. It doesn’t hadnle sub second response times, various timezones, or even high uptime requirements. If it breaks once a month and you nudge the agent to fix it, that’s fine. If the UI is ugly but everyone knows where the button is, that’s fine. If it only handles your group’s data shape and nobody else’s, that’s the point.

Once you drop the requirement for production-grade engineering, the return on investment flips. An hour of prompting that saves fifteen minutes a day—or untangles a recurring coordination mess among friends—is a compounding win. A weekend of traditional development for the same payoff never was. AI doesn’t just make building faster; it makes low-stakes building rational.

What comes next

We’re going to see a lot more tiny utilities, personal automations, shared dashboards for friend groups, scripts dressed up just enough to click. Not because the industry decided small software was a category, but because individuals finally have a way to chip away at the backlog of private inconveniences that used to be permanently deferred.

The interesting shift isn’t that software gets smaller. It’s that the unit of valuable software stops needing to be a product. Sometimes the highest leverage thing you can build is something a small number of people will ever use.