Geocode the places
The agent reads the entities in your knowledge graph and geocodes every address and place name into coordinates. It never invents a coordinate — if something is too ambiguous to resolve, it says so instead of guessing.
Map Canvas
Geocoding lifts the addresses and places already in your graph onto a map, and the agent authors an interactive map for each question. Hotspots, by-region rollups, reachability, routes — and a feature you click leads straight back to its node in the knowledge graph.
From the places in your graph to a thematic map — geocoded, authored, and linked back to the graph.
The agent reads the entities in your knowledge graph and geocodes every address and place name into coordinates. It never invents a coordinate — if something is too ambiguous to resolve, it says so instead of guessing.
It picks the right layers — scatter, heatmap, choropleth, routes — and computes new spatial facts like hotspots, rollups, and reachability with map transforms. You author it by asking, in plain words.
Each mapped place remembers its location back in the knowledge graph, so you can ask spatial questions over the whole workspace — and click any feature to pivot to its graph node.
Turns place names and addresses into coordinates, and turns coordinates back into the administrative region they fall in.
Heatmaps and hex bins surface where things concentrate, so a flat list of points reads as a pattern.
Rolls points up into districts and shades them by a rate — a point-in-polygon aggregation, not a guess.
Real travel-time catchments, real road routes, and nearest-by-time — coverage and access you can actually trust.
A time scrubber replays how hotspots and regions shift — incidents, shipments, rollouts, as they evolve.
Every coordinate, distance, and class break is computed by libraries and services, never by the model. The map stays honest.
Local-first. The map is authored from your own knowledge graph, and a feature you click leads straight back to its node.