Building Stellara: The First Space LLM

“I want to be in the room.” That was the response I gave to a soon-to-be mentor and friend when he asked me what I wanted to do. I was fresh out of graduate school and sitting at a table across from a person who would go on to help me throughout my career, including now. What I was communicating was that I wanted to be in the room where the big decisions were being made. The room where we just heard about some major problem and the people inside were charged with solving the problem. No playbook. No checklist. Just a tough problem and smart people. I wanted to be on that team. About 15 years after that conversation, I found myself in one of those rooms. It was during the COVID-19 pandemic so the physical room I was in sat inside a secure area of DHS Headquarters. I was in that room alone as the only representative from DHS on the line. I sat in front of a camera, as so many grew accustomed during this period, with colleagues from the White House and government departments around the beltway. We were there in the opening hours of Russia’s invasion of Ukraine. Top of the agenda? The Russian hack of American company Viasat in the opening hours of the invasion. No playbook. No checklist. Just a tough problem. I was in the room. Subscribe now What I remember most from that situation was not that first meeting but the months that followed. I remember when we couldn’t discuss the issue publicly, then I remember when we could. I remember hearing friends from the State Department use clearly approved talking points to say the right things with the right tone. I also remember the effort that went into figuring out what had been affected by this action. From my chair at DHS, this meant critical infrastructure. Did this outage have an impact on any critical infrastructure sectors? What about national critical functions? And you’ll recall this was an attack on the ground segment of Viasat, not on the vehicle itself. As many smart analysts and technicians figured out the extent of the damage, I was left thinking about what would have happened if an incident like targeted US interests. What might have been if there was an attack on an orbital vehicle with the express and malicious intent to disrupt critical services in the US homeland? Were we ready for something like that? It was another two years before I found myself sitting in the cool weather inside a screened-in back porch in Florida in January. It was the home of my wife’s parents, and we were in town for a few weeks. I had departed the government about 9 months before to cofound Frontier Foundry. We were enjoying some early success, but the satellite issue continued. Now with a team focused on the issue, we considered things like the complexity of low earth orbit (LEO), interdependencies in our critical infrastructure, and the ongoing war in Ukraine, I came up with a thesis. Leaving aside the space as critical infrastructure debate for a moment, if we were ever going to understand the risks from the complexities emerging in LEO, we would have to be able to know what the risk of an outage of an individual satellite is. From that thesis emerged other questions such as knowing how many satellites support a given critical infrastructure sector. Without understanding these first principles, the debates we were having around space cybersecurity, critical infrastructure, and risk were missing critical data points. The idea for Stellara was born on a cool night in Central Florida. Share Our team spent weeks combing through publicly available data and creating a very rough spreadsheet draft. The idea was to prove, even if it was ugly, that it was possible to connect individual satellite missions to critical infrastructure missions. The earliest drafts are never something I’d show publicly, but they were important pieces of the creative process. It wasn’t perfect but it showed promise. There were many steps left but the foundation was there, and we could all see a path toward something special. In the months that followed, our incredible tech team made my awful early spreadsheets into an AI-driven engine. They used that output to create the product that we now call Stellara. Stellara now has an LLM front end that lets users ask direct questions about real or exercise scenarios in space. “What are the primary satellites that impact the emergency services sector?” “What US critical infrastructure sectors are impacted by an outage of the INMARSAT 3 F5 satellite?” What it became was a way to issue spot before the incident happened. It was a way to see around the corner and anticipate vulnerabilities. As LEO gets more crowded, Stellara became the answer to understanding risks to our critical infrastructure. And we wouldn’t have to learn the lessons the hard way. Stories about a tech company bringing a product to market are not hard to come by. Companies identify market gaps, find product market fit, and build new products daily. But the story of the conceptualization and creation of Stellara speaks not only to identifying a national and homeland security problem and building for it, but also the culture that Frontier Foundry is building. We know what we build, but the Stellara story speaks to why we build. One of the original marketing lines we used for the company was “Today was Once the Unimaginable Future.” The line refers to many of the biggest problems we face today that were once viewed as out of our technical reach. Each day we innovate, that line is truer, and more problems come within our reach. But what made the Stellara effort special wasn’t that a tech company identified a problem it could solve. What made it special was that we gave our builders and innovators the latitude they needed. The first version, actually the first few versions, were ugly and they weren’t to the level of accuracy we needed. But we have a culture that demands that people be willing to take risks and come forward with a product that is not finished but shows the concept. Stellara solves a once impossible problem: how to determine what critical infrastructure sector is impacted by an outage of an individual satellite . It solves that problem by applying multi-modal, privacy preserving AI in unique ways on top of unique data. The secret sauce is the willingness to look at a once unsolvable problem and believe in yourself and the team around you to solve it. Lessons abound in a startup as any founder will tell you. Good and bad lessons came from the Stellara work. We did some things well and other things less well. As we are building more products, we are bringing the same energy and willingness to do the hardest work on the hardest problems. Hard problems without playbooks or checklists will always demand the creativity of smart people. At Frontier Foundry, I’m thankful we built a team exactly like that. Connect with us: LinkedIn , Bluesky , X , Website To learn more about the services we offer, please visit our product page. Nick Reese is the cofounder and COO of Frontier Foundry and an adjunct professor of emerging technology at NYU. He is a veteran and a former US government policymaker on cyber and technology issues. Visit his LinkedIn here . This post was edited by Thomas Morin, Marketing Analyst at Frontier Foundry. View his Substack here and his LinkedIn here . Leave a comment