Spatial Stack with Matt Forrest

What does it actually take for an AI to navigate you through the real world, without a screen and without a camera running the whole time?
In this episode of the Spatial Stack, Matt Forrest sits down with Sean Gorman and Pramukta Rao, the co-founders of Zephr, to unpack how they ground large language models in location. Sean and Pramukta spent twenty years building startups together, from GeoIQ to Snap's visual positioning work, and at Zephr they walked away from the camera and went back to the sensors already in your phone.
They get into why the blue dot on a map breaks down the moment you stop looking at a screen, why language models are bad at spatial reasoning like left, right, and across the street, and how they fix it by doing the geometry first and handing the model clean language. Instead of training ever-bigger foundation models, they push a small model and well-structured data down to the edge so the conversation stays fast.
In this episode, we cover:
- Why they left visual positioning and AR cameras behind after Snap
- Cooperative positioning: getting survey-grade accuracy out of commodity phones
- GNSS and the urban canyon problem (3 meters to 50 meters and back)
- Beyond the blue dot: building a first-person, egocentric experience
- Why LLMs struggle with geometry, and what context engineering solves
- Small models at the edge vs giant foundation models
- Overture Maps, GERS IDs, and conflating POIs with imagery to "agree on reality"
- The grounding service: MCP, REST, and Opus or Gemma on device
- Conversations about place as a new geospatial primitive
Whether you build with spatial data, work on AI navigation, or just want to see where location and LLMs are heading, this conversation is your field guide.
Connect with Zephr:
Website: https://zephr.xyz
Sean Gorman (LinkedIn): https://www.linkedin.com/in/sean-gorman-93a79
Pramukta Rao (LinkedIn): https://www.linkedin.com/in/pramukta/
Zephr (LinkedIn): https://www.linkedin.com/company/zephr-xyz
00:00:00 – Intro and twenty years of startups together
00:04:05 – Why they walked away from cameras and AR at Snap
00:04:54 – Commodity sensors and cooperative positioning
00:07:15 – GNSS 101 and the urban canyon problem
00:10:56 – Beyond the blue dot: an egocentric experience
00:13:37 – Why LLMs are bad at left, right, and across the street
00:18:19 – Context engineering and the retrieval problem
00:22:03 – Small models at the edge vs giant foundation models
00:30:09 – Collective memory: OpenStreetMap, Mapillary, Overture
00:32:08 – Conflating POIs with imagery to agree on reality
00:35:36 – The grounding service: MCP, REST, Opus or Gemma on device
00:38:37 – What's next: conversations as a new geospatial primitive
00:42:22 – Where to find Zephr
📊 FREE: The Modern GIS Skill Map
The 5 skills that actually matter in modern GIS (and what you can stop learning). Based on a survey of 1,400+ geospatial professionals.
➡ Get the free training + PDF guide: https://forrest.nyc/go/training/
🚀 Join The Spatial Lab:
Stop guessing at your career path. Get direct mentorship, advanced training, and a roadmap to these high-value roles inside The Spatial Lab.
👉 https://forrest.nyc/spatial-lab/
📰 Daily modern GIS insights: https://forrest.nyc
CONNECT WITH ME
📸 Instagram:  https://www.instagram.com/matt_forrest/
💼 LinkedIn: https://www.linkedin.com/in/mbforr/
📧 Newsletter: https://forrest.nyc
🌐 Website: https://forrest.nyc

In this episode of the Spatial Stack, Matt sits down solo to work through a fight that broke out across Reddit and LinkedIn last week: is GIS dying, or is it just changing its name?
It started with a GIS manager who argued the field is alive, pointing to a friend who landed a $200,000 job building autonomous systems, a role where GIS never appeared in the title or the description. The comments pushed back. If the keyword for our whole profession barely returns jobs anymore, is the field really growing?
Matt's answer is that both sides are right. The work of spatial is expanding into data engineering, software, and product roles. The GIS title is contracting at the same time. It's one trend seen from two directions, the same way geology departments quietly became geoscience without the work ever changing.
He makes the case that the title was never the skill. Spatial intuition is, and it's the thing that transfers into higher-paying roles that don't carry the GIS label. Then he closes with a challenge: describe who you are and what you do without using the words GIS, spatial, or geospatial.
Whether you're job hunting, stuck in the technician trap, or an employer who wants this skill set but doesn't know what to call it, this conversation is about how we position the work going forward.
---
📊 FREE: The Modern GIS Skill Map
The 5 skills that actually matter in modern GIS (and what you can stop learning). Based on a survey of 1,400+ geospatial professionals.
➡ Get the free training + PDF guide: https://forrest.nyc/go/training/
CHAPTERS:
00:00:00 – The Reddit post that started it
00:01:00 – Two views: the work is growing vs. the title is shrinking
00:02:34 – Why "stay positive" doesn't pay the rent
00:03:33 – Spatial is special: what actually transfers
00:04:53 – The geology-to-geoscience analogy
00:05:54 – The title of GIS was never the skill
00:07:09 – The technician trap and where the salaries moved
00:08:18 – Listener comments: spatial judgment, the map as interface
00:10:32 – What to do now: SQL, Python, and cloud-native beyond the toolbox
00:11:51 – Building a portfolio on LinkedIn
00:13:22 – The Spatial Intuition Challenge
00:15:32 – Put your face on it: make it a video
CONNECT WITH ME
📸 Instagram:  https://www.instagram.com/matt_forrest/
💼 LinkedIn: https://www.linkedin.com/in/mbforr/
📧 Newsletter: https://forrest.nyc
🌐 Website: https://forrest.nyc

What does it actually take to make AI useful on location data?
In this episode, Matt Forrest sits down with Ryan Urabe, co-founder and CTO of Dataplor, to unpack how AI, embeddings, and agents are changing the way we work with points of interest and places data.
Ryan explains why general-purpose models already understand spatial concepts but still struggle to execute them, and why the real unlock is the harness around the model, not a geospatial-specific model. He walks through Dataplor's data-quality philosophy, the category problem (why "supermarket" and "grocery store" have zero string similarity but near-zero conceptual distance), and how embeddings let them measure conceptual distance across 10^9 places and even across languages.
Whether you build with spatial data, lead a data team adopting AI, or you are trying to figure out what embeddings actually do, this conversation maps out what is working today and what is still forming.
In this episode, we cover:
- Why AI is an accelerant for data quality, not a replacement for it
- Treating AI like a capable employee on their first day
- Where general models fall short on spatial problems
- DuckDB as the Swiss Army knife for orchestrating spatial data
- The category problem and conceptual vs. semantic distance
- How embeddings map 7-Eleven in Tokyo and Tennessee to the same concept
- A vision for agentic AI built natively for geospatial
- The "end of the scarcity of intelligence" framing for where this is all heading
Connect with Ryan:
LinkedIn: https://www.linkedin.com/in/rurabe/
Website: https://www.dataplor.com
Email: ryan@dataplor.com
LEARN MORE
Dataplor's agentic SaaS product is launching this summer. To start your complimentary trial, contact: freetrial@dataplor.com
📊 FREE: The Modern GIS Skill Map
The 5 skills that actually matter in modern GIS (and what you can stop learning). Based on a survey of 1,400+ geospatial professionals.
➡ Get the free training + PDF guide: https://forrest.nyc/go/training/
00:00:00 – Cold open
00:01:01 – Welcome and Ryan's background
00:03:44 – Why AI still struggles with location
00:05:13 – Bringing Dataplor's data into AI, product and team
00:08:36 – How technical teams are adopting AI
00:10:10 – Treating AI like a capable new employee
00:12:20 – Where general models fall short on spatial
00:16:44 – DuckDB and opinionated workflows
00:18:14 – Data quality as the whole game
00:20:58 – The category problem: supermarket vs. grocery store
00:27:53 – Embeddings and conceptual space, with a 3D walkthrough
00:38:22 – A vision for agentic AI in geospatial
00:43:37 – The end of the scarcity of intelligence
00:47:25 – Where to find Ryan and Dataplor
📰 Daily modern GIS insights: https://forrest.nyc
CONNECT WITH ME
📸 Instagram:  https://www.instagram.com/matt_forrest/
💼 LinkedIn: https://www.linkedin.com/in/mbforr/
📧 Newsletter: https://forrest.nyc
🌐 Website: https://forrest.nyc

What does it actually take to map every agricultural field on Earth?
In this episode, Matt sits down with Jen Marcus, Vice President of Strategic Innovation Programs at Taylor Geospatial, and Isaac Corley, Director of AI/ML Research at Taylor Geospatial and a torchgeo maintainer, the team behind Fields of The World (FTW).
In late April they released the first globally consistent dataset of agricultural field boundaries, at 10m resolution, fully open on Source Cooperative. They dive deep into how it came together, from building the fiboa format to standardize ground truth across 24 countries, to running model inference across the entire planet, to shipping it with a confidence layer instead of pretending it was perfect. You'll hear honest perspective on what GeoAI can really do today and where the hype outpaces reality.
In this episode, we cover:
- Why a global field boundary map had never been done, and why no single organization was positioned to do it
- The labeled-data problem and why models have to generalize to places like South America and Africa with little ground truth
- The fiboa format and Chris Holmes's "architectures of participation"
- How the Technical Fellows program turned open-source contributors into the core team
- Running global inference efficiently with Sentinel-2 planting and harvest mosaics
- Cloud-native outputs (GeoParquet, PMTiles, Zarr) you can stream with no backend
- What's real vs. what's marketing in geospatial AI, and the ImageNet lesson
- What's next: stakeholder feedback loops, higher-resolution imagery, and mapping new features beyond fields
Whether you build ML pipelines, work with satellite data, or you've ever wondered how much of the planet is still genuinely unmapped, this conversation breaks it down without the buzzwords.
LINKS:
Fields of The World: https://fieldsofthe.world
Dataset on Source Cooperative: https://source.coop/wherobots/fields-of-the-world
Taylor Geospatial: https://taylorgeospatial.org
Jen Marcus
LinkedIn: https://www.linkedin.com/in/jennifer-marcus-b559091/
Isaac Corley
Website: https://isaac.earth
LinkedIn: https://www.linkedin.com/in/isaaccorley/
GitHub: https://github.com/isaaccorley

📊 FREE: The Modern GIS Skill Map
The 5 skills that actually matter in modern GIS (and what you can stop learning). Based on a survey of 1,400+ geospatial professionals.
➡ Get the free training + PDF guide: https://forrest.nyc/go/training/
CHAPTERS:
00:00:00 – Cold Open
00:01:01 – Welcome and Guest Intros (Jen Marcus and Isaac Corley)
00:02:32 – Why Map Field Boundaries, and Why It Had Never Been Done
00:05:53 – Going from Local to Global Scale
00:07:19 – Architectures of Participation and the fiboa Format
00:12:44 – The First St. Louis Meeting and the Technical Fellows Program
00:18:04 – Running Global Inference at Scale
00:22:54 – Cloud-Native Outputs on Source Cooperative
00:25:05 – Why This Matters and What's Real vs. Hype
00:28:39 – The ImageNet Lesson and Holding a North Star
00:32:20 – What's Next for Fields of The World
00:36:12 – Impact, an OpenStreetMap for Fields, and How to Get Involved
00:40:33 – Postdoc, Tech Fellows, and Looking Out the Airplane Window
🚀 Join The Spatial Lab:
Stop guessing at your career path. Get direct mentorship, advanced training, and a roadmap to these high-value roles inside The Spatial Lab.
👉 https://forrest.nyc/spatial-lab/
📰 Daily modern GIS insights: https://forrest.nyc
CONNECT WITH ME
📸 Instagram:  https://www.instagram.com/matt_forrest/
💼 LinkedIn: https://www.linkedin.com/in/mbforr/
📧 Newsletter: https://forrest.nyc
🌐 Website: https://forrest.nyc

In this episode of the Spatial Stack, Matt sits down with Troy Schmidt, a 20-year GIS developer and the creator of SPHERE, an open-source Python package that runs FEMA's HAZUS flood risk methodology on GeoParquet and DuckDB.
Troy dives into why depth damage functions are simpler than the engineering language suggests, and how the gap between the HAZUS methodology and the HAZUS software pushed him to build a Python-first alternative. He shares the moment he realized the data and the science were already public, and that the only thing missing was a modern stack to run it on.
The conversation also covers the three types of flooding most people lump together (coastal, riverine, and pluvial), why pluvial risk is the gap that nobody insures, and what an open core model means for geospatial science.
Finally, Troy walks through his cloud-native discovery process, from Wherobots and Earthmover webinars to DuckDB and vectorized math, and explains why turning legacy methodology into a Python package unlocks deployment patterns that were never possible before.
📊 FREE: The Modern GIS Skill Map
The 5 skills that actually matter in modern GIS (and what you can stop learning). Based on a survey of 1,400+ geospatial professionals.
➡ Get the free training + PDF guide: https://forrest.nyc/go/training/
Connect with Troy Schmidt:
LinkedIn: https://www.linkedin.com/in/mr-troy-schmidt/
SPHERE on GitHub: https://github.com/Niyam-Projects/sphere
CHAPTERS:
00:00:00 – Intro
00:01:35 – Welcome and Troy's 20-Year GIS Journey
00:04:17 – The Risk Modeling Landscape
00:07:25 – Coastal, Riverine, and Pluvial Flooding
00:11:05 – HAZUS Methodology vs. HAZUS Software
00:16:05 – Site-Specific vs. Census Block Analysis
00:18:05 – Building SPHERE: A Python Package for Risk
00:21:25 – Cloud-Native, GeoParquet, and DuckDB
00:25:05 – Why the Methodology Was Simple All Along
00:27:05 – Discovery: Webinars, Wherobots, and the Modern Stack
00:29:35 – Spatial as a Boundary Data Type
00:32:35 – The Open Core Model and the Bridge to Modern GIS
00:38:05 – Ensemble Models and What's Next for Climate Risk
00:39:35 – Where to Find Troy and SPHERE
🚀 Join The Spatial Lab:
Stop guessing at your career path. Get direct mentorship, advanced training, and a roadmap to these high-value roles inside The Spatial Lab.
👉 https://forrest.nyc/spatial-lab/
📰 Daily modern GIS insights: https://forrest.nyc
CONNECT WITH ME
📸 Instagram:  https://www.instagram.com/matt_forrest/
💼 LinkedIn: https://www.linkedin.com/in/mbforr/
📧 Newsletter: https://forrest.nyc
🌐 Website: https://forrest.nyc