Business context
The client is a GIS platform provider serving urban planning and geospatial users. The platform already held rich GIS data and analytics, but getting answers from it took GIS skills. That limited how many people in a customer's organization could get value from the product.
The challenge
The client wanted users to get geospatial insights without building GIS queries themselves. Plain language questions had to become correct spatial operations on real data. The AI layer needed structured access to the existing PostGIS data and cloud services, and answers had to appear in the product users already know.
Our solution
We built an MCP server that exposes the platform's existing GIS data and analytics as tools an AI model can call. We also built a conversational interface in the product UI, connected to the GIS backend.
- MCP server over GIS data and analytics, built with FastMCP
- Conversational interface in the UI, based on LobeChat and React
- Integration with PostGIS, AWS AppSync and AWS S3
Business impact
- Business users who do not write GIS queries can ask the platform questions and act on the answers
- GIS specialists spend less time on routine requests from colleagues
- The client adds an AI feature to its product that builds on data and analytics it already has
- The MCP server opens the platform to AI agents and assistants beyond its own UI
