GTA DataCity - San Francisco co-working dataset visualized, as a game!

A tiny 3D platformer atlas of San Francisco's AI coworking spaces.

Add a comment

Replies

Best
Introduction: GTA DataCity is a n x m table presented as a game Explanation: Running around a city is far more fun than reading a table Inspiration: As an explorer, I love discovering the world. As a nerd, I often spend hours researching before the trip and multiple spreadsheets are often involved. Here is my SF co-wokring dataset ,shared as a game. Evolution: The game was birthed in a Stanford x Deepmind hackathon and launched shortly after. The list has been many months in making. CTA: Contribute by adding/raving about your fav spots in the city
How did GPT-5.6 change the ambition or scope of what you shipped?
I turned my research into a playable game. An n x m table was expressed as a 3D world game, all because of GPT 5.6. This product was envisioned in a Stanford hackathon, after which my agent worked for 9.5hrs to get me the first version the next morning. No other AI model works for that long or that well. This build is particularly interesting because I do not know ANYTHING about game building - so GPT-5.6 and I talked in fundamentals and it executed in code. None of this was possible before

pretty cool. would be hype if you make it so that I can go inside the building and see the vc offices lol

 thanks, getting in VC offices is hard lol.

With this launch, I'm focused on getting as many new data points for the game directory (hence, SF co-working directory) as possible. The game is just a fun visualisation for this research :)

love that research turned into something you'd actually want to click through instead of a spreadsheet. since it's open source and the schema is basically "n x m table -> playable city," is the pipeline generic enough that someone could point it at a different city's dataset and get their own version, or is SF specifically wired into the game logic right now

This is the most fun anyone has made a dataset this year. Turning a list of co-working spaces into a place you walk around means I actually retained where things sit relative to each other, which a table has never once done for me. Please do more cities. Any plan to open the format so people can drop in their own dataset?