Why we let AI generate map layouts instead of asking users to

Every map-building tool eventually presents the same screen: a blank pin, a color picker, and thirty settings. What shows in the popup? Which field is the title? Should the rating be a badge or a line of text? The tool is asking you to be an information designer, when all you did was show up with a spreadsheet.

We think that screen is the reason most embedded maps look the way they do.

Your columns already are a design brief

A spreadsheet of places is not unstructured data. Column names carry intent: Name, Type, Price range, Best for, Website. A row that has a photo URL wants a card with an image zone. A column of one-to-five numbers is a rating and wants to render like one. A Notes column full of sentences is body copy, not a badge.

Sightfield's meta-model reads those columns and makes the calls a designer would make: which field leads, which details earn a labeled row, which values become pills, what the card hierarchy is. You get a designed layout on the first upload — not a blank canvas.

The result, before any settings

A live map will appear here

This showcase is waiting for its published map.

Adjust anything, redesign nothing

AI picks the starting point; it does not lock the door. Every zone of the card is editable — swap what leads, hide a field, override a single place's layout. The difference is where your time goes. You spend it on the last five percent that is genuinely yours (brand color, tone, one stubborn card), not on the first ninety-five percent that is pattern recognition.

The pitch was never "AI makes maps." It is: the output quality of a designed guide, at the effort cost of an upload.