GeoSports

GeoSports is a daily sports geography game built around a deceptively simple idea: every sports question is answered by dropping a pin on a map. You get five questions a day — the ballpark the Chicago Cubs have called home since 1916, the city whose team was blocked from trading for Chris Paul in 2011, the birthplace of a champion — and the closer your pin lands to the real location, the more points you score. A round takes about a minute, needs no email address or account, and ends with a result designed to be shared with friends. It is the kind of daily habit that fits between two meetings and still leaves you feeling like you learned something.

The game was created by sports media creator Frank Michael Smith, who says he knew almost nothing about software engineering when he opened Claude Code on a weekend in May 2026. Twelve hours and hundreds of back-and-forth prompts later, GeoSports was live. It had 79 players on day one, 575 on day two, and more than 4,000 before noon on day three after a sports newsletter featured it. By the end of May it had reached one million unique users, and by September it had passed 15 million rounds played. The audience was not a niche AI crowd; it was sports fans who liked a daily ritual and a shareable score.

What makes GeoSports work is not the code generation. Smith used AI to build a known format around a subject he already understood deeply. He still writes the questions himself, because, as he told Sportico, AI does not understand the pulse of the sports fan. The game now has a Pro tier at $5 a month that unlocks past days, random rounds and sport-specific modes, and it has since expanded into a GeoGames app alongside GeoHistory and GeoFooty.

For anyone building with AI, GeoSports is a clean example of the right division of labor. The model handled the map, the scoring, the interface, the daily reset and the referral loop. The human supplied the taste: which questions are fun, which locations are satisfying to find, and why a sports fan would come back tomorrow. That is why a twelve-hour project turned into a real business while thousands of technically impressive demos did not.