Location Insights for Your Graphs with Neo4j, PostGIS, and Brickgraph
summary
Demonstrates a production flow that pairs Neo4j for graph objects with PostGIS (and PGVector) for geometry and embeddings so the system supports semantic plus geospatial search and can query millions of geometries on the database side. Geocoding uses Mapbox to get lat/lon, PostGIS executes ST_DWithin/KNN radius searches and a travel‑time service runs concurrently with a Cypher graph query. A FastAPI + SQLModel backend feeds a Kensington agent (Pyantic AI + Anthropic Son of 4) that returns LLM text plus structured metadata for map visualizations.
AI-written summary · about
chapters (11)
- 0:00Introduction to location insights
- 1:29Demo of Kensington tool
- 3:15Why use geospatial analytics
- 6:23PostGIS overview and advantages
- 9:17Integrating PostGIS with Neo4j
- 12:09Simple geospatial search with PostGIS
- 14:38Complex search and retrieval workflow
- 15:20What is Brickgraph
- 17:27Building agent tools with Pydantic AI
- 20:19Kensington agent workflow
- 21:29Resources and closing
related talks
26mBuilding Geospatial GraphQL APIs By Leveraging GraphQL Database Integrations
55mLet Me Graph That For You
51mTime to graph up with Spring Data Neo4j
32mNever use a graph database!
22mQuerying & Visualizing Graphs in Postgres with Apache AGE
46mGetting Down To Earth: Geospatial Analysis With Rails
43mLocate all the things
53mSpatial Programming for the Rubyist