22-25 April 2026 
Making Location Intelligence AI-Native, From Spatial Queries to Natural Language
Proposed session for SQLBits 2026TL; DR
Discover how Databricks, Microsoft Fabric, and DuckDB can be combined with OpenAI to make location intelligence AI-native. See how natural language interfaces can generate and explain spatial queries, enabling broader access to geospatial analytics while remaining grounded in enterprise data platforms.
Session Details
Geospatial analytics has traditionally required specialist skills, complex SQL, and a deep understanding of spatial data types. As a result, location intelligence often remains siloed within technical teams, limiting its wider adoption across the business.
In this session, we explore how modern data platforms such as Databricks, Microsoft Fabric, and DuckDB are enabling scalable spatial analytics, and how the addition of AI-driven, natural language interfaces is fundamentally changing how users interact with location data. By combining cloud-scale spatial processing with large language models such as OpenAI, we can move from writing complex spatial queries to simply asking questions like, “Which areas are underserved?”, “Where is risk increasing?”, or “What locations should we prioritise next?”.
Through practical demos and real-world examples, we will show how spatial queries can be generated, explained, and validated using AI, reducing the barrier to entry for non-specialists while preserving the rigour and performance required for enterprise analytics. The session will focus on grounded, production-ready patterns rather than speculative AI, highlighting where these approaches work well and where human oversight remains essential.
You will come away with a clear understanding of how to design AI-native location intelligence solutions that sit naturally within your existing data platform, enabling broader adoption of geospatial analytics without sacrificing control, governance, or performance.
What will we cover?
- Why geospatial analytics remains inaccessible to many users, despite modern data platforms
- How Databricks, Microsoft Fabric, and DuckDB support scalable spatial processing using SQL and columnar data formats
- The role of open standards and shared data, including GeoParquet, OpenStreetMap, and Overture Maps
- Architectural patterns for combining spatial engines with AI services safely and effectively
- Using OpenAI to translate natural language questions into spatial SQL queries
- Techniques for explaining, validating, and auditing AI-generated spatial logic
- AI-assisted exploration of large spatial datasets without requiring deep GIS expertise
- Real-world examples of AI-enhanced location intelligence across industry use cases
- Live demos showing natural language to spatial query workflows in Databricks and Power BI
- Best practices for governance, security, and cost control when introducing AI into spatial analytics
In this session, we explore how modern data platforms such as Databricks, Microsoft Fabric, and DuckDB are enabling scalable spatial analytics, and how the addition of AI-driven, natural language interfaces is fundamentally changing how users interact with location data. By combining cloud-scale spatial processing with large language models such as OpenAI, we can move from writing complex spatial queries to simply asking questions like, “Which areas are underserved?”, “Where is risk increasing?”, or “What locations should we prioritise next?”.
Through practical demos and real-world examples, we will show how spatial queries can be generated, explained, and validated using AI, reducing the barrier to entry for non-specialists while preserving the rigour and performance required for enterprise analytics. The session will focus on grounded, production-ready patterns rather than speculative AI, highlighting where these approaches work well and where human oversight remains essential.
You will come away with a clear understanding of how to design AI-native location intelligence solutions that sit naturally within your existing data platform, enabling broader adoption of geospatial analytics without sacrificing control, governance, or performance.
What will we cover?
- Why geospatial analytics remains inaccessible to many users, despite modern data platforms
- How Databricks, Microsoft Fabric, and DuckDB support scalable spatial processing using SQL and columnar data formats
- The role of open standards and shared data, including GeoParquet, OpenStreetMap, and Overture Maps
- Architectural patterns for combining spatial engines with AI services safely and effectively
- Using OpenAI to translate natural language questions into spatial SQL queries
- Techniques for explaining, validating, and auditing AI-generated spatial logic
- AI-assisted exploration of large spatial datasets without requiring deep GIS expertise
- Real-world examples of AI-enhanced location intelligence across industry use cases
- Live demos showing natural language to spatial query workflows in Databricks and Power BI
- Best practices for governance, security, and cost control when introducing AI into spatial analytics
3 things you'll get out of this session
Understand why natural language to SQL struggles with spatial analytics, and how to address it
Learn how to translate business questions into scalable, validated spatial SQL using Databricks, Fabric, DuckDB, and OpenAI
Take away practical architectural patterns for introducing AI into location intelligence safely and effectively