Generating Running Routes with GPT-6 Astra and ChatGPT Work

A user used the ChatGPT Work interface and the GPT-6 Astra model to generate 5 km and 10 km looped running routes starting from their own address. They requested the use of OpenStreetMap data, and received the generated routes as visualizations and as downloadable GPX and GeoJSON files. The process took about 27 minutes and the result matched the user’s request exactly. Both the 5 km example route and the 10 km version were produced as loops returning to the starting address. The solution demonstrates that large-model tools can process geospatial data and produce exportable spatial data files, which can be useful for amateur and professional runners, urban planners, and mapping applications. It also raises privacy and security concerns because the user provided their exact home address. The file formats (GPX, GeoJSON) and the embedded visualization make sharing and tracking the routes easier. The example highlights the practical applicability of generative AI for location-specific tasks and suggests that the efficiency and automatic export options can support training planning and personalized route design.