Beyond Chat: Building an Agentic AI Swarm for National Security
The rapid advancements in artificial intelligence present a pivotal moment for national security. However, the focus on chatbot-style interfaces obscures a more significant possibility: leveraging AI to create autonomous “agent swarms” capable of accelerating decision-making and overwhelming adversaries. the lessons from the conflict in Ukraine, coupled with a pragmatic approach to integrating commercial and open-source AI models, offer a path forward for the United States.
Currently,much of the discussion around AI in defense centers on individual “bots” and the debate between open-source and commercial solutions. This misses a crucial point. The key to avoiding vendor lock-in isn’t choosing one over the other, but rather owning the orchestration layer - the control plane – that manages and audits interactions with these models. As Ben Van Roo, CEO of Legion Intelligence, argues, ”The antidote to vendor lock-in is owning the orchestration layer and audit trail. Control the interfaces to control the leverage.” This control plane must incorporate robust security features,including level entitlements,lineage tags,and a critical “kill switch” for rapid intervention.
This architecture allows for a flexible approach, routing tasks to the most appropriate model – commercial for sophisticated reasoning, open-source for handling sensitive, classified data on air-gapped networks. Crucially, any contracts with AI vendors should mandate open APIs to prevent future integration challenges. this shifts the focus from a ”bot gallery” to a connective tissue for smart agents. Lessons learned from initiatives like NIPRGPT can be incorporated into this control plane, creating a unified system capable of operating across diverse AI services.
The Ukrainian conflict vividly demonstrated the power of saturation – overwhelming an adversary with volume to disrupt their decision-making process. America’s response should be to field an “agent swarm” designed to accelerate U.S. decision loops. This isn’t about replacing human analysts, but augmenting their capabilities.The distinction between AI ”assistants” and “agents” is critical: assistants merely speed up tasks like typing,while agents autonomously complete missions,report results,and proceed to the next objective,fundamentally altering operational tempo.
Consider the example of security clearance updates. Currently, analysts spend significant time manually cross-referencing databases and completing paperwork. an agent swarm could automate this process, monitoring personnel records for triggering events, automatically gathering documentation, pre-filling forms with cited sources, routing approvals based on reviewer availability, and tracking completion metrics. This frees analysts to focus on exception handling and quality control, dramatically increasing efficiency. This pattern can be replicated across numerous administrative functions.
The necessary technology already exists. The challenge isn’t innovation, but implementation. As Van Roo emphasizes, ”What’s missing is the will to move beyond pilots to production.” While debates over contract terms and security architectures continue, potential adversaries are rapidly compressing their decision cycles. The choice is clear: prioritize fielding an operational agent corps or remain focused on admiring the potential of chat interfaces. In the digital battlespace, as in Ukraine, controlling tempo dictates outcomes.
The agent swarm is inevitable. The question facing the United States is whether it will be the one launching it, or defending against it.
Ben Van Roo is the co-founder and CEO of Legion Intelligence, an agentic AI platform built for the Department of Defense. Ben has spent his career building tech companies serving the public and private sectors, including five years as researcher at RAND and a startup executive for 13 years. Ben has a PhD in operations research from the University of Wisconsin-Madison.
Please note, as a matter of house style *War on the Rocks will not use a different name for the U.S. department of Defense until and unless the name is changed by statute by the U.S. Congress.*
