Which of our assets are within 5km of a flood zone and had an incident last quarter? That's a five second question and a forty minute query. For a decade in Oil & Gas I watched great spatial questions die inside that gap.

BigQuery has had serious geospatial functions for years and almost nobody uses them. Gemini is what finally makes them reachable.

In this session we take a synthetic fleet telematics dataset, we land it in BigQuery and we build a layer where you ask questions in plain language and get back both an answer and a map. Synthetic on purpose, so everything you see on stage is something you can rebuild on your own data on Monday. We'll explore the geospatial functions that really pay off, how to ground the model so it writes correct spatial SQL instead of plausible nonsense, how to keep costs sane when the geography gets big, and how to render results so a non technical stakeholder gets it immediately.

If your data has coordinates in it, and most data does, you are sitting on questions you never thought to ask!

Your data has coordinates. So why do you never ask spatial questions? Because writing the SQL costs more than the answer is worth! Gemini plus BigQuery changes that. Live demo: you ask in plain language, the map answers. Geospatial functions worth knowing, prompts grounded so the SQL is actually correct, costs kept sane. Come and discover the questions you never thought to ask!