131. Edge AI for the Warfighter with Jason Rathje, President of Public Sector at webAI
All Quiet on the Second FrontShow Notes
Most AI being built for the military assumes a steady connection to a data center. In a contested fight, that connection is one of the first things to go. Jason Rathje has spent a decade building new ways for commercial companies and capital to reach national security, and now he’s building AI that runs on the devices soldiers already carry. He joins Tyler to talk about designing AI from the edge up, and what the defense tech market looks like after ten years of change. What's happening on the second front: Why local-first architecture wins when power, latency, and connectivity are the real constraints How small, mission-tuned models for comms, intel, and fires can work together like a fire team What a peer-to-peer AI mesh makes possible when there’s no central service to call back to Why talent follows capital, and how new government financing tools changed the math on hard tech The case for keeping dual use, critical minerals, and the industrial base in the national security conversation
Jason is President of Public Sector at webAI and the founding director of the DOW’s Office of Strategic Capital, and previously ran AFWERX’s AFVentures program after six years as an Air Force engineer.
More from All Quiet on the Second Front
Harvard offers a course called Doing Business with China. As Nick Maynes points out, there is no course for doing business with America. Meanwhile, students across Harvard and MIT are building sensors, AI, and manufacturing tech with real national security applications, and most of them have never thought of the government as a customer or a partner. Nick and Ryan Holte started QLab to close that distance. They join Tyler Sweatt to talk about finding dual-use founders before those founders know they are dual-use, funding them without taking equity, and teaching technical builders how to sell to the government and to investors. Why QLab takes zero equity and funds founders with non-dilutive grants so they build full time instead of taking a summer internship The first perception barrier every student founder has to get past, and what national security companies actually look like: cyber, manufacturing, critical minerals How to keep one demanding early customer from pulling a product into a niche that won’t scale Why technical founders get about 30 seconds to make an investor curious, and how to translate SBIR wins into de-risking Building for America across administrations, and why national security has to stay bipartisan
Buying satellite imagery used to take a year of contract negotiations, and the images still arrived as PDFs. SkyFi launched in 2023 to simplify the process and had users in 180 countries within two weeks. By the time government customers came calling, the product was already battle-tested by commercial users, ready for a market that rarely gives second chances. This week, Luke Fischer, a 20-year Army veteran, former special ops aviator, and now SkyFi co-founder, joins Sam Gray to share what he's learned building a dual-use company commercial-first. Leveraging your commercial base to show up ready for the gov Why a continuing resolution can make a dual-use company more attractive to investors What happens when warfighters, not leadership, do the selling Staying competitive with AI as a small company
Roughly $100 billion a year goes into federal IT, and by GAO’s count about 80% of it goes to maintaining old, complex legacy code bases. As AI speeds up everything built around them, those code bases become the thing the whole enterprise waits on. Jake Sortor ran into this himself as an intelligence officer in Iraq, trying to get a Python script he'd written tested on a secure network. He leads public sector at Blitzy now, and his argument is that AI is the only thing that attacks a backlog this size — with humans still making every decision that matters.
In 2020, telling people that organized actors were using bots to shape what the public believes got you called a conspiracy theorist.
Twenty years ago, a young logistics lieutenant solved an impossible field problem with a Franklin Planner, a satellite phone, and a contact who happened to be at his desk. It worked — but it ran on luck. Bill Pessin has spent his career since making sure the mission never has to. He joins Tyler Sweatt to trace the line from Army logistics to leading Salesforce’s Missionforce, and why acceleration, not just speed, is the point.