GenAI Automates Space Force Data Discovery

Generative artificial intelligence (GenAI) is helping the Space Force automate data search and discovery options. The agency is leveraging Amazon Web Services’ Marketplace for the U.S. Intelligence Community (ICMP) to quickly deploy these innovative solutions, Space Force Senior Technical Program Manager Will Haskell said Monday at AWS re:Invent in Las Vegas, Nevada.

ICMP allows users to streamline procurement and provisioning of third-party software and services in air-gapped environments.

“Traditional software preparedness is slow and cumbersome, ill suited to the rapid pace of the cloud and AI development,” said AWS ICMP lead Bryanna Tucci. “By the time you get that software solution or that GenAI model, it’ll be outdated. … Our public sector customers, our national security missions and our citizenry, deserve [better].”

Haskell said ICMP puts the end user closer to the acquisitions process than what was previously possible through traditional procurement methods, enabling Space Force to leverage new solutions to solve for longstanding challenges, like data search and discovery.

“We put a pretty large burden on analysts and Intelligence Community users to know exactly where data … resides,” said Haskell. “There are tons of data. We don’t have a data problem, but we do have a data discovery problem. And, unless you know where that tool is or where that data set is, good luck finding [it].”

Haskell said the Space Force is using GenAI to gather data from returned imaging. The GenAI tool, GURU, is helping the Space Force improve data efficiency and accessibility. The newest feature allows users to ask general questions about an image, like “how many vehicles are present?”

“As a signals analyst, I don’t really know what’s interesting about an overhead shot of some barracks,” Haskell told GovCIO Media & Research following his panel. “But I could ask what do you see in this image? And it might spit out something that piques my interest, that correlates to another piece of intelligence.”

Haskell explained that within the Intelligence Community, the same data is often stored in multiple locations. GURU addresses this by working directly with data where it resides. It combines traditional APIs with vectorization—a technique that enhances data processing by performing multiple operations simultaneously on a set of data points. This enables GURU to provide responses about unstructured data, such as returned imaging, more efficiently.

“We are bringing all of those disparate data sources together to formulate an overall AI generated response, potentially pointing users to data sources or tools that they did not know about previously,” said Haskell.