Taylor Geospatial Institute (TGI) and Amazon Web Services (AWS) continue to drive innovation through its Generative AI Challenge. On March 6, 2025, TGI hosted its latest GeoAI Working Group meeting, which highlighted yet another round of practical applications of generative AI – how it is reshaping industries, enabling smarter and more proactive decision-making, and changing the way we work with geospatial data.

Marge Cole, Director of Strategic Initiatives, opened the discussions by highlighting the transformative role of Artificial Intelligence (AI) in geospatial applications. “We are at a tipping point where AI is not just an enhancement to geospatial technology, but it’s a fundamental shift in how we understand and interact with data,” she said. 

Cole also highlighted the importance of collaboration – working alongside innovators, researchers, government entities, and industry leaders – to drive advancements in AI for geospatial applications. This collaborative approach ensures that AI’s potential is not merely explored but turned into actionable solutions that make a real-world impact.

Her remarks set the stage for the presentations from the next round of TGI AWS Generative AI Challenge winners – Sparkgeo, Bedrock Research, and Tera Analytics – each presenting their AI-driven solutions that are pushing the boundaries of geospatial intelligence. From advancing environmental conservation to reimagining how we analyze satellite data and predict ecological risks, these companies are pioneering AI applications in geospatial data to address complex global challenges.

[Don’t miss out on the next TGI GeoAI Working Group meeting scheduled for April 3. Click to Register]

James Banting, Vice President of Research at Sparkgeo, delivered the first presentation, as he shared how their AI-driven solution is automating Biodiversity Net Gain (BNG) assessments for land developers. BNG is a UK government policy that ensures any land development must leave habitats in a better state than they were found. This policy is mandatory for developers in the UK, who are required to demonstrate a 10% improvement in biodiversity.

Banting explained the challenges developers face when applying these policies, particularly with the complex, spatially-intensive data involved. “The UK government has Excel sheets that you fill out as part of your assessment for land development. This is fantastic for a language model, as we can train something to look at this and start interpreting it with the right geospatial context,” he said, while highlighting the need for a system that can not only interpret the documents but also apply geospatial context, such as the British National Grid, which is a key reference for spatial data in the UK.

Banting’s team is leveraging Large Language Models (LLMs) to automate the interpretation of these complex documents and help developers meet BNG standards more efficiently. “There’s some special sauce we can bring to the table to make sure that LLMs know a little bit about geospatial context,” he said, referring to how Sparkgeo plans to use geospatial context to enhance the accuracy of these models and make the system more user-friendly for developers.

By combining AI with spatial data and policy documents, Sparkgeo’s solution aims to streamline the process of ensuring land developments meet environmental and regulatory standards while also improving biodiversity.

Matt Reisman, Co-Founder and CTO of Bedrock Research, talked about how AI is transforming the way satellite data is processed. Bedrock Research is developing cross-modal foundation models that allow for seamless integration and analysis of different data types – satellite imagery, sensor data, and even rada – to derive comprehensive insights from multiple sources in real time. This is crucial for quick decision-making, especially in situations involving security or disaster management.

“AI is changing the way we understand data and allowing us to act on it more intelligently and with greater precision, especially in critical situations,” he emphasized.

Reisman was joined by Peter Shagnea of AgileView, which is pushing the envelope with synthetic data creation, and Mark Bowersox of Urban Sky, which is sharing impactful imagery captured from high-altitude balloons. These contributions highlight new frontiers in high-resolution remote sensing and offer a fresh and innovative approach to data collection.

Bedrock Research’s cross-modal AI models integrate a variety of data sources – from SAR to visible and infrared satellite data – which are combined to create a more complete picture of the environment and provide decision-makers with the ability to understand complex situations quickly and accurately.

The final presentation was delivered by Robert Carroll, Co-Founder and CEO of Tera Analytics, who discussed the role of predictive modeling in shaping sustainable environmental decisions. Tera Analytics is integrating generative AI into its workflows to simulate future environmental scenarios using satellite imagery, drone data, and climate predictions to help businesses and governments plan and adapt proactively rather than reactively.

Carrol showcased the interesting Property Risk Index developed by Tera Analytics, which uses satellite data and predictive modeling to assess environmental risks on properties. “By simulating future environmental scenarios, we are giving organizations the insights they need to make decisions that not only react to change but proactively shape a more sustainable future,” he said. The Property Risk Index enables organizations to prepare for risks such as flooding, wildfires, or other environmental events. 

This ability to simulate various “what-if” scenarios focuses on prevention and proactive action and allows for better preparedness, whether it’s predicting the effects of climate change on agricultural yield or understanding the long-term impact of deforestation. 

Future of AI in Geospatial Decision-Making

From automating environmental risk assessments to enabling real-time analysis of satellite imagery, the meeting emphasized how AI is transforming the way we collect, analyze, and act on geospatial information for use in a wide range of critical applications.

Marge Cole’s closing remarks captured the essence of this transformation: “AI is driving a new era of decision-making – one that is faster, smarter, and more sustainable.” 

The next TGI GeoAI Working Group Meeting will take place online on April 3, 2025, from 2–4 pm CDT, where TGI will showcase the next round of Generative AI Challenge winners, including:

• Christie Capper of Deep Earth — Geothermal Atlas for Adoption and Energy Transitions — 2.25 pm
• Sean Gorman of Zephr - GPS-Location Enabled GenAI — 2:45 PM
• Hamad Alemohammad of Clark University - Models for Land Cover Changes — 3:15 PM
• Danny Sheehan of Crosswalk - Create Intelligent Agents Enabling Emissions Models — 3:45 PM

Click here to register for the April 3 meeting.

About Taylor Geospatial Institute 

TGI is passionate about fueling geospatial science and technology to create the next generation of solutions and policies that the whole world will depend on for sustainability and growth. 

The TGI consortium includes Saint Louis University, the Donald Danforth Plant Science Center, Harris-Stowe State University, University of Illinois Urbana-Champaign, Missouri University of Science & Technology, University of Missouri-Columbia, University of Missouri-St. Louis, and Washington University in St. Louis. Collectively, these institutions cover geospatial research from ocean depths to outer space.  

For more information, visit taylorgeospatial.org.