Simon Willison Builds Map Viewer Using GPT-5.6-Sol

Prominent developer Simon Willison has released a new interactive GeoJSON Map Viewer, a utility designed to help users visualize geographic data on an OpenStreetMap interface. The tool allows users to paste GeoJSON objects, such as Feature, FeatureCollection, or Geometry, directly into an editor. From there, users can customize styling options like fill color and opacity, render the features on the map, and export the final visualization as a PNG image.

The creation of the tool highlights a highly collaborative workflow involving multiple cutting-edge artificial intelligence models. Willison initially sought recommendations for mapping utilities from GPT-5.6-Sol, which proactively generated the foundational codebase for the tool. To refine the application, Willison used Claude Code for web-based iterations and completed the final version with the assistance of Fable 5.1.

Beyond the development of the viewer itself, Willison highlighted how modern AI models simplify the acquisition of complex spatial data. To populate the map, he used ChatGPT Work to gather precise local political boundaries for the Granada Community Services District and the Midcoast Community Council near Half Moon Bay, California. The AI successfully extracted and merged disparate government data sources to generate the necessary GeoJSON files.

For software practitioners and data analysts, this development showcases a shift toward highly automated, multi-model development pipelines. Instead of manually writing boilerplate code or wrestling with complex GIS software, developers can now leverage specialized AI agents to both write the visualization software and compile the underlying datasets. The integration of models like GPT-5.6-Sol and Claude Code suggests that the barrier to creating bespoke, single-purpose data tools has virtually disappeared.