AI Creates Custom Running Routes on Demand

A new application using GPT-6 Astra (Max) and ChatGPT Work demonstrates the power of generative AI for practical applications. Users can now request personalized running routes that are generated in real-time, complete with interactive maps and downloadable files.

The process works by providing a simple prompt like “I live at [my address]. Figure out 5K and 10K running routes from me that loop from my house. Use OSM data.” The AI then locates the user’s position using Nominatim and Overpass, calculating optimal routes based on OpenStreetMap data.

Interactive Visualizations and Downloadable Files

The generated routes are presented as embedded visualizations within the ChatGPT interface, showing clear path overlays on interactive maps. Users can also download the route data in standard formats like GPX and GeoJSON for use with fitness trackers or other mapping applications.

When asked about its methodology, the AI explained: “I used Nominatim to locate the address and Overpass to download local OpenStreetMap roads and trails, then calculated the loops locally.” This showcases how LLMs can leverage external tools to provide complex solutions beyond their internal knowledge base.

The application also highlights a challenge with current LLM systems—the inability to retrieve past conversations or code after they’ve been compacted. While the initial route generation was successful, requesting the underlying Python code proved impossible once the thread had been processed.