
Spatial Stack with Matt Forrest #34: Everything Is Changing in Geospatial, Here’s What Actually Matters
Dec 17, 2025
Explore the transformative changes in geospatial technology over the past year. Discover where AI truly enhances workflows without eliminating jobs. Learn how cloud-native geospatial is now in production systems. Delve into groundbreaking formats like GeoParquet and their impact on data analytics. Understand why systems architecture is becoming vital, and how 'power skills' like collaboration are essential in modern roles. Get insights on the emerging dynamics within technical and leadership positions in spatial teams.
AI Snips
Chapters
Transcript
Episode notes
Geospatial Embeddings: Big Leap, Real Limits
- Embeddings and large geospatial models like Alpha Earth advanced discovery but have limits.
- They help find similar areas and test hypotheses but can't replace object detection or forecasting yet.
Use Language Models For Specific Tasks
- Use language models where they fit: summarizing, cleaning, and small agentic tasks work well.
- Build agentic workflows and small models for concrete tasks like data cleaning and code snippets.
Shift From Tools To Systems Thinking
- The conversation shifted from single tools to systems thinking and how tools fit together.
- Matt emphasizes combining strengths of formats like GeoParquet, PostGIS, and DuckDB into coherent architectures.
