There are roughly 200 petabytes of Earth observation data floating around in orbit, according to LGND AI’s own marketing [LGND.ai, retrieved 2026]. That is a lot of pictures. It is also, for most businesses, a completely useless pile of pictures. A satellite image of a forest is just a picture until you can ask it if there’s a wildfire starting, or if a new road has appeared where it shouldn’t be. LGND AI, a Philadelphia-based startup founded in 2024, is betting that the key to unlocking that value isn't a better camera, but a better index.
They are building what they call geospatial-AI infrastructure: a system that ingests satellite and aerial imagery and converts it into vector embeddings. These are compact numerical representations that turn a visual landscape into something an AI model can search, compare, and reason about. The company offers both an enterprise application for analysts and an API for developers, aiming to make this data accessible for tasks like risk modeling, change detection, and logistics planning [Perplexity Sonar Pro Brief].
The Bet on a Searchable Planet
The core bet is that the future of geospatial analysis is not in manual inspection, but in querying. Instead of a human staring at before-and-after photos, an insurance underwriter could ask an AI agent to “show me all properties within 500 meters of areas with increased wildfire risk scores over the last month.” The agent would query LGND’s vector index,dubbed the Strabo index in company materials,to find the relevant embeddings and return an answer [LinkedIn, Arek Romanski, retrieved 2026].
This shifts the unit of value from the raw terabyte of imagery to the actionable insight. For sectors like insurance, finance, and logistics, where location-based risk and opportunity are everything, that translation could be worth quite a lot. The company’s $9 million seed round, led by Javelin Venture Partners in July 2025, suggests investors think so too [PR Newswire, July 2025].
A Team Built for the Mission
The founding team reads like a shortlist of people who have spent careers trying to make planetary-scale data useful. CEO Nathaniel Manning was the first Chief Data Officer at USAID, a White House Presidential Innovation Fellow, and, most notably, a co-founder of Kettle, an AI-driven climate reinsurance company [Perplexity Sonar Pro Brief]. He knows the specific pain of pricing risk in a changing climate and the data gaps that make it hard.
His co-founders bring complementary depth. Dan Hammer serves as chief product officer, and Bruno Sánchez-Andrade Nuño, the chief scientist, is credited with founding the Microsoft Planetary Computer initiative [brunosan.eu, retrieved 2026]. The broader team pulls from Google, NASA, the World Bank, and Mapbox, a mix that suggests equal parts AI engineering, scientific rigor, and geospatial product sense [Perplexity Sonar Pro Brief].
| Founder | Role | Key Background |
|---|---|---|
| Nathaniel Manning | CEO | Co-founder of Kettle (climate reinsurance), former USAID Chief Data Officer [Perplexity Sonar Pro Brief] |
| Dan Hammer | Chief Product Officer | Co-founder, former CEO at Parken Sport & Entertainment [Bloomberg Markets, retrieved 2026] |
| Bruno Sánchez-Andrade Nuño | Chief Scientist | Founder of Microsoft Planetary Computer, co-founder of Clay [brunosan.eu, retrieved 2026] |
Traction and Technical Scale
Public customer names are scarce, but the company has cited pilot work with professional-services firms and logistics companies [Perplexity Sonar Pro Brief]. More concrete is the technical scaling story. According to a case study from Union AI, LGND scaled its geospatial AI workloads from zero to 160 GPUs to process satellite imagery and build its embeddings [Union AI, September 2026]. That kind of compute footprint isn’t for a demo; it’s for building a foundational data layer.
The company’s early positioning targets a wide swath of enterprise:
- Insurance and Finance: For dynamic risk assessment of assets based on environmental changes.
- Logistics and Supply Chain: For monitoring infrastructure and optimizing routes.
- Government and Climate Adaptation: For tracking deforestation, illegal mining, or urban development [Perplexity Sonar Pro Brief].
The Incumbent It Must Beat
The most obvious counterfactual isn’t another startup, but the status quo: in-house data science teams building custom pipelines. A large insurer or agribusiness could, in theory, hire a team of ML engineers and geospatial analysts, procure satellite data feeds, and build their own embedding models. It would be expensive, slow, and require rare expertise, but it’s the default for many who need this data today.
LGND’s answer is that its platform abstracts that complexity into an API. The bet is that the cost and speed advantage of a unified, maintained service will outweigh the perceived control of a bespoke solution. The risk is that the most valuable customers,those with the budget and need,may still choose to own the core capability, viewing geospatial intelligence as too strategic to outsource.
The Unit Economics of a Planetary Layer
Let’s run a back of the envelope calculation. If LGND’s infrastructure can help a single large reinsurer improve its wildfire risk models by even a fraction of a percentage point, the financial impact is measured in tens of millions of dollars annually, easily dwarfing a software subscription. The real test will be proving that their embeddings generate insights proprietary data teams cannot. The company’s path to becoming essential lies in becoming the de facto translation layer between the physical Earth and the digital decisions of enterprise AI. To win, they don’t need to replace every analyst; they need to become the map every AI agent checks first.
Sources
- [LGND.ai, retrieved 2026] Company Website | https://lgnd.ai
- [Perplexity Sonar Pro Brief] LGND AI Startup Brief
- [LinkedIn, Arek Romanski, retrieved 2026] Post on Strabo vector index
- [PR Newswire, July 2025] LGND Raises $9M to Make Earth Data Intuitive and Actionable | https://www.prnewswire.com/news-releases/lgnd-raises-9m-to-make-earth-data-intuitive-and-actionable-302502527.html
- [brunosan.eu, retrieved 2026] Bruno Sánchez-Andrade Nuño Background | https://brunosan.eu
- [Bloomberg Markets, retrieved 2026] Dan Hammer Profile | https://www.bloomberg.com/profile/person/16299616
- [Union AI, September 2026] LGND scales geospatial AI from zero to 160 GPUs with Union AI | https://www.union.ai/case-study/lgnd-scales-geospatial-ai-from-zero-to-160-gpus-with-union-ai
- [TechCrunch, July 2025] LGND wants to make ChatGPT for the Earth | https://techcrunch.com/2025/07/10/lgnd-wants-to-make-chatgpt-for-the-earth/