The app that kept living

The Reisi work finally has a falsifiable shape: a functional demo where you ask the agent for things and it does them, except the money is fictional and nothing connects to a real rail ▸ 1:14. Add analytics, put people in front of it, and the core supposition, that people prefer telling an agent what to do with their money over hunting through menus, becomes measurable ▸ 1:42. The audience for those numbers already exists: the family-side investor who asked for evidence, plus a cold email he’s now willing to send ▸ 2:37. The padrino, for the record, never replied, which still puzzles him, the fix was only “mejorar un poquito la oferta” ▸ 2:57.

los usuarios llegaron solos; imagínate si alguien cuidara la puerta →

The night’s second discovery vindicates the Divo principle from the other direction. Koby, the Kobo-highlights product, was improved with Claude, released open source, and left for dead, and its analytics say it refused to die: around thirty monthly users, seven weekly ▸ 3:36, thirty-nine registered, and while he sat watching the dashboard, a login arrived from Italy ▸ 4:04, with most usage coming from the United States ▸ 4:18. Not a business yet, “pero siempre es alguito”: an app that acquires users with zero marketing while its maker isn’t looking is the strongest organic signal anything on the LuarAI shelf has produced. The ganas arrive on cue, then meet the pile of accumulated bugs ▸ 4:26, which is its own honest metric of what abandonment costs.

Two products, one lesson pointing both ways: validation is manufactured for Reisi because no one has asked for it yet, and was free for Koby because someone already had…