Speed is often treated as the primary virtue in technology deployment, including in discussions about AI and government modernization. Justin Fulcher disagrees, at least about where the emphasis should fall. In a piece published by IT Security Guru, the former Defense Department advisor and technology founder makes the case that durability, not deployment velocity, is the right standard for evaluating public-sector AI.
The argument connects to his broader diagnosis of what is wrong with government technology programs. Agencies are not slow because they lack resources or motivation. They are slow because decades of outdated processes, siloed data systems, and compliance rules designed for paper-based workflows have created what Justin Fulcher calls institutional drag. The drag is compounding. Each workaround generates new workarounds. Each legacy system constrains what can be built around it.
What His Pentagon Work Demonstrated
Fulcher speaks from experience. As a Senior Advisor to the Secretary of Defense, he worked on acquisition reform and technology modernization, contributing to efforts that cut software procurement timelines from years to months. That kind of result does not happen from launching fast and iterating in public. It comes from understanding the institutional constraints in advance and designing around them from the start.
He co-founded RingMD, a telemedicine platform that operated across Asia, before his government work. Building healthcare technology across multiple regulatory environments taught a similar lesson: systems that endure are those designed with institutional constraints in mind, not those designed to disrupt and then seek forgiveness.
A Practical Standard
Justin Fulcher’s framework for evaluating AI tools is practical. Tools that require extensive retraining, introduce new compliance concerns, or create fresh failure points will struggle to gain traction in government, regardless of how impressive they look in a demonstration. The tools that work are those that reduce existing friction. Measured against that standard, many AI products currently marketed to public agencies would not pass. That reality, more than any policy debate, shapes how Fulcher approaches the question of AI in government modernization. Refer to this article for related information.