Getting artificial intelligence into a federal agency is one thing. Getting people to actually trust it is another. Justin Fulcher, a technology founder and former Pentagon advisor, argues that the second challenge is the one agencies tend to underestimate, and it’s the one that determines whether AI tools survive past their pilot phase.
Stewardship Over Speed
“Serious work is defined less by certainty at the outset than by stewardship over time,” Fulcher wrote in a LinkedIn article on public service and responsibility. It’s a line that reflects his broader philosophy: technology adoption in government isn’t about moving fast and breaking things. It’s about building systems that hold up under scrutiny for years, not just during a demo.
That perspective comes from firsthand experience. Fulcher served as a senior advisor to the Secretary of Defense, working on acquisition reform, and earlier co-founded RingMD, a telemedicine company operating across Asia. Both roles put him in regulated environments where trust had to be earned before adoption could happen at scale.
Design Constraints, Not Afterthoughts
For AI tools to succeed in government, Justin Fulcher argues they need to be auditable and explainable from the start, capable of failing safely rather than breaking silently. That requirement isn’t a nice-to-have. Data security rules, civil service protections, and public accountability standards make it a baseline expectation for anything deployed at scale.
He also points out that tools requiring heavy retraining or introducing new compliance headaches rarely last, no matter how impressive their underlying capabilities. The ones that integrate cleanly into existing workflows, and visibly save time without adding complexity, are the ones that earn continued use. Fulcher’s takeaway is straightforward: durability beats novelty every time in institutions built to last.
That kind of patience runs counter to how most new technology gets marketed, with an emphasis on speed and disruption over reliability. Fulcher’s advice to agencies weighing AI tools is simple: judge them the way you’d judge any system meant to operate for a decade, not by how it performs in a single demonstration. Refer to this article to learn more.
See Justin Fulcher Facebook https://www.facebook.com/JustinLFulcher/