There’s a defense company built entirely around one uncomfortable fact.
Most AI tools need the internet to work. Most soldiers on the ground do not have the internet to give them.
Ian Kalin runs a company called TurbineOne. He served in the Navy as a counter-terrorism officer and a nuclear engineer before he built anything. He said something on a recent Federal News Network interview that deserves more attention than it’s getting.
He said close to 80 percent of national security professionals can’t reach cloud-hosted AI tools in the environments where they actually work. Too remote. Too contested. Too dangerous to light up a phone and pull a signal. That’s Kalin’s own figure, from his own company’s experience building for that gap, because a number like that should come with whose mouth it came from attached.
But the number isn’t the part of the interview that matters most here.
Terry Gerton, the host, asked him a direct question. As AI becomes more common across military operations, what should commanders be most cautious about?
Kalin’s answer didn’t dodge it.
Every new system, he said, gets treated as higher risk the moment it shows up. Not because commanders are stubborn. Because the system already in place has a track record, and the new one doesn’t yet. Trust gets built with consistency over time. There’s no shortcut around that, and there shouldn’t be one.
TurbineOne celebrates that skepticism. They don’t treat it as an obstacle to route around. They treat it as the correct starting posture for anyone being asked to hand real decisions to a machine they haven’t watched perform yet.
That is the whole argument this Baseline has been making from day one, said by a man who has never read a word of it.
Trust that’s mandated from the top down is brittle. It complies exactly to the edge of the order and stops there. Trust that’s earned by consistent performance, watched and tested in the actual conditions where it has to hold up, is a different animal. It doesn’t need to be forced, because it’s already proven itself.
Kalin made a second point on its own.
Gerton asked him directly — with all this new AI capability arriving, is there a risk of leaning on it too far? Automating too many decisions?
His answer: you don’t lose the old requirements when you add new ones. He pointed to the war in Ukraine as proof — old-world trench warfare and next-generation drone warfare, running side by side, at the same time, on the same ground. Neither replaced the other. The new capability sits on top of what was already true, not in place of it.
That’s the same shape as a standard this framework has carried from early on — that governance built for AI operates above a floor it cannot reach or replace. New tools don’t retire the old requirements for judgment, caution, and human accountability. They add a layer. They don’t erase the one underneath.
Put the two points together and you get something close to a complete picture of how trust in AI is actually supposed to work, whether you’re commanding troops or running a company.
Don’t ask for blind faith. Earn trust the slow way, through consistency, out where it can be tested and watched. And when the new capability arrives, don’t assume it replaces what came before it. Assume it stacks on top, and the old requirements are still live underneath it.
Kalin ended the interview on a line that fits this whole conversation better than he probably knew. The industry is moving fast, he said. So is the pace of innovation ahead. But the thing worth remembering through all of it is that the people matter more than the hardware.
A battlefield commander figured that out from hard experience, not from a governance framework.
That’s exactly the kind of convergence this Baseline was built to watch for.
This post was drafted with AI governed assistance and reviewed and directed by Michael S. Faust Sr. before publication.
Contact: micvicfaust@gmail.com
© 2026 The Faust Baseline LLC | All Rights Reserved
