In this talk we will look at how PostGIS and Uber's H3 index can be used for aggregating large amounts of data, in our case property insurance risk, in real-time. We will explore a number of different techniques from the H3 PostGIS extension generating GeoJSON, to generating MVTs from the database to pre-caching the H3 index and painting a vector tile layer client side. For our client side layer will use a React JS interface, Maplibre and will also look at Deck.GL for more advanced use cases. We will discuss how the stack can be deployed using a serverless architecture running on AWS Lambda and Aurora Serverless Postgres. This talk requires no prior knowledge however some experience with PostGIS and vector tiles will be useful. You will learn techniques which can be applied to any problem domain where there is the need to work with data volumes where processing individual points would not be practical.