Models training on models of NYC — no photos, no GPS

Peter B. is showing how you can teach an AI to reason about New York City without feeding it a single photo or street-level image. The trick is to build a structural map of the city — streets, blocks, parks, lines — and then use a vision model to draw that map as a sketch. The model sees the drawing, reads the geometry, and learns the layout. No OpenStreetMap dumps, no satellite tiles. Just shapes and labels.

The reason this matters is that most geospatial AI still leans on heavy imagery: satellite feeds, LiDAR, street-level photos. Those are expensive, privacy-heavy, and slow to update. A line drawing of the city is cheap to generate, easy to version, and doesn't carry the same surveillance baggage. For anyone building routing, delivery, or neighborhood tools, it's a lighter-weight alternative.

The post walks through the pipeline: a structural representation, a vision model rendering it, and the model learning to read what it drew. It's a proof-of-concept, not a production system — but it shows one way to skip the image-heavy stack for city-level reasoning.

Why this matters for us: lighter, sketch-based maps could make neighborhood tools faster and cheaper for delivery routes, gig work, and local logistics — without piling on more satellite surveillance.