@joulee - Hyperpersonalized Software
@joulee on X The author, a parent and developer who runs many AI agents at once, argues that software is moving from one-size-fits-all products toward personal tools people can reshape for their own workflows and tastes. She describes building her own agent dashboard and a math and reading app for her kids as proof that building for yourself is now cheap enough to be worth doing.
- Building a custom agent interface took her about half an hour to get a working version, and she built a mobile and desktop app that combines Claude and Codex, which neither official app supports.
- Personalized math stories tied to her kids' real lives reportedly help engagement, and she cites a 2013 study of 145 ninth graders where personalized algebra problems improved speed and accuracy, especially for struggling students.
- Settings menus rarely get used: a Jared Spool study found fewer than 5% of people changed a single Microsoft Word setting, which she uses to argue that fixed rule-book customization falls short.
The author, a parent who runs many AI agents from the command line, argues that personalization has finally become cheap enough to make software one-of-one. Her opening frame is Andy Warhol's line that a Coke is a Coke no matter how much money you have, which Marc Andreessen echoed in 2023 by adding browsers, smartphones, and chatbots. Her counterpoint is that software used to be too expensive to build for anyone but a mass audience, so everyone got the same settings-menu-level customization. She cites Jared Spool finding that fewer than 5% of several hundred people had changed a single Microsoft Word setting, and Clay Shirky's 2004 claim that building for a few dozen users was an "absurd target population." Her thesis is that this no longer holds: building tailored software is now cheap enough to be worth doing for yourself.
Her main example is Mog Squad, a mobile and desktop client she built to control her eight terminal windows of AI agents. She had been envious of her husband's polished remote access to Claude and Codex, but after about half an hour with an AI coding tool she had a mostly working version. Her version lets her use both Claude and Codex, which the official apps don't, and it shows the number of active subagents as sparkles, puts the status verbs back in her own wording, and arranges her eight agents in a grid. She settled on eight because fewer makes her restless and more makes her lose track of projects. She also invokes the IKEA effect, the tendency to value things you built yourself, to explain why she now has little desire for third-party tools.
Her second example is a math and reading app for her kids, modeled on Duolingo. Instead of gamification like streaks and multiplier boosts, the lessons use stories drawn from the kids' real lives, friends, hobbies (fashion and theater for her daughter, Minecraft, Geometry Dash, and Pokemon for her son), and family trips, with a few family values slipped into what the characters do. The kids can request stories through a feedback button, and she made holographic collectible cards of their stuffed animals as rewards. She backs the approach with a 2013 study of 145 ninth graders, where students given algebra problems rewritten around their interests solved them faster and more accurately, with the biggest gains for the struggling students and effects that persisted after the personalization was removed. Her broader claim is that personalized software beats uniform software for the individual, and that the cost barrier that once justified uniformity is gone.