Pick a tool, then click the map.
Ready for analysis
A fresh street graph is built from OpenStreetMap for the selected location.
Loading San Francisco as the opening example…
GeoPulse builds a street network for whatever place you search, pulls it live from OpenStreetMap, and analyses what that network reaches. You choose a travel mode and a radius, and it returns the reachable area, the shape of the street grid, and the points of interest that fall inside it. Nothing is precomputed and no city list is baked in: the analysis runs against current OpenStreetMap data at the moment you ask for it.
Travel mode matters more than most people expect. A network that looks dense when driving often thins out sharply on foot, because footways, crossings, and pedestrian links are tagged separately from roads. Switching between drive, walk, and bike on the same location is usually the fastest way to see where a place has been built for cars and where it has not.
The 15-minute city is the proposition that daily needs — food, school, health care, work, and green space — should sit within a short walk or cycle of home. As an urban planning slogan it is contested. As a measurement it is useful, because it converts a vague question about neighbourhood quality into something you can check: within this radius, on foot, what is actually there.
A walkability figure is a summary, and summaries hide things. Two neighbourhoods can score alike while one is bounded by a motorway nobody can cross and the other is not. Severance, gradient, lighting, and whether the footway exists in usable condition are not in the data. Treat the score as a prompt to look closer rather than as a verdict.
OpenStreetMap is the only global street dataset that is openly licensed, editable, and current enough to answer a question about an arbitrary place. That openness is also the limitation: coverage is uneven and reflects who has mapped an area. Dense European and North American cities are richly tagged; many cities in lower-income countries are mapped well for roads and thinly for footways and amenities. A low score in an under-mapped area may describe the map rather than the place, which is the same distinction that governs every data system built on volunteered information.
GeoPulse is one of the tools in a wider portfolio of AI and data systems: see the full list of builds and the AI projects written up in more depth. The reasoning behind building systems that hold up in places with unreliable infrastructure is set out in the Last-Mile AI Framework, and the same data-quality questions raised above run through humanitarian AI.
Built by Shahzad Asghar. Map data from OpenStreetMap contributors.