9-7-2026 –, Brunsemazaal Taal: English
Off-the-shelf network-analysis tools force a choice: bend your research question to fit the software, or wrestle with heavyweight server stacks. Agentic AI changes that calculus. Instead of adapting to someone else's schema, I could specify the exact data models, request formats, and features my work demanded and have a working tool emerge in days rather than months.
The result is NetWeevil: an open-source engine for reproducible network analysis that runs entirely on your own machine, directly against OpenStreetMap and GTFS data. From a single OSM extract it builds the street network, compiles routing profiles (walk, bike, car), and answers routing, matrix, service-area, and accessibility queries in milliseconds. A GTFS module adds multimodal transit access, and a companion QGIS plugin pulls routes, isochrones, and accessibility surfaces straight onto the canvas. Not the first to do this, but first to do like I want it to based on tens of ideas and questions about network analysis over the years.
This talk is two stories in one: how agentic AI reshapes who gets to build geospatial software and how fast, and a tour of the tool that process produced. Attendees will leave with a concrete picture of AI-assisted prototyping and a practical, private engine for multimodal network analysis on their own data.
