Army Reserve Commander Teaches GenAI for Public Affairs Operations

U.S. Army Reserve Lt. Col. Raymond Ragan stood before public affairs specialists at Fort Dix, New Jersey, on April 9, 2026. He wasn't teaching tactical maneuvers or field communications. He was teaching them how to prompt an artificial intelligence system to write news articles faster than their adversaries could spread disinformation.

The class covered GenAI, the Department of War's bespoke AI platform, and how to use it within the constraints of military policy. Ragan commands the 361st Theater Public Affairs Support Element, a unit under the 99th Readiness Division that deploys mobile public affairs detachments worldwide. The training focused on three core areas: rapid article production, effective prompt engineering, and ethical boundaries for AI use in military communications.

"We are in a cognitive fight, told through the narrative to support multidomain operations, and the enemy can lie," Ragan told the room. "As the old adage says, a lie can travel halfway around the world while the truth is putting its shoes on, and that is a challenge for public affairs."

The physical reality of this work involves clicking through multiple verification systems, cross-referencing operational security guidelines, and ensuring every word complies with current policy. All of which takes time. Ragan's point was simple: speed matters when adversaries are already publishing false narratives before the Army can verify facts.

Ragan holds two AI patents while also serving as a chief information officer for a startup company that develops frictionless customer service AI in his civilian capacity. This dual expertise informed his approach to the five-part prompt structure he demonstrated during the session.

First, you give the system a persona: a Defense Information School-trained Soldier. Then you provide context: what information you have on the subject. Next, you state the ask: I want you to create something. After that, you specify constraints: write it according to AP style, but don't use hyphens or semicolons. Finally, tell it to interview you to clarify before generating a response. This last step is critical because the AI might ask questions about things you hadn't thought of (which is actually the point of the exercise).

Ragan ran various prompts through GenAI to demonstrate the process and discuss the product produced. He highlighted the differences between general prompts and the five-part prompt, pointing to the strengths and weaknesses of each output. The soldiers watching could see the cursor blinking, the text appearing line by line, the tangible difference between a vague request and a structured one.

U.S. Army Reserve Staff Sgt. Tyler Matz, a public affairs mass communications specialist with the 326th MPAD, arrived skeptical. "Before this class, I've always looked at AI as something I don't fully understand and do not want to use," Matz said. The demonstration changed his view. He now plans to practice with GenAI and incorporate the techniques into training at his unit.

While using GenAI is critical to increasing the speed at which public affairs products are produced, Ragan emphasized that leaders cannot pass responsibility on to the system. Nothing leaves the newsroom that hasn't been personally signed off by someone. The Army wants soldiers to use AI, and it is authorized for use in public affairs, but there are hard constraints on it.

To further emphasize the point, Ragan covered key parts of Department of Defense Instruction 5400.19, Public Affairs Use of Artificial Intelligence. The document highlights guiding principles, policy limitations, and transparency guidelines for generative AI use. U.S. Army Reserve Pfc. Isabella Youngblood, a public affairs mass communication specialist with the 354th MPAD, appreciated learning both the limits and possibilities. She plans to use GenAI to brainstorm article angles and draft approaches, understanding that it serves as a starting point, not a replacement for her judgment.

The training also addressed the Generative AI Decision Matrix in the DODI, which clarifies when operators need to notify the public that generative AI has been used in content creation. This transparency requirement adds another layer to the workflow—soldiers must track not just what they write, but how they wrote it.

Ragan wrapped up the class emphasizing the information environment that the Army is currently in. "Our adversaries are moving much more quickly, so we need to move much more quickly," he said. "We must use AI as a tool to get content published quickly to refute and counter an enemy narrative."

He closed with a perspective on where the technology is headed. "No matter how bad the AI is today, it is the worse AI you will ever use. In other words, it is always going to get better." The implication for operations leaders is clear: understanding how to work with these tools now positions teams to adapt as the capabilities improve.

Whether this training translates into measurable improvements in narrative speed remains to be seen. The real test comes when units deploy and face actual information operations in contested environments. For now, the 361st TPASE has at least given its soldiers a framework to work with.

The class itself was documented on the official Army website, which serves as the primary record of the event. Independent coverage from completeaitraining.com corroborates the timeline and scope of the training.

Whether users actually pay attention to the five-part prompt structure in the heat of operations is another question entirely. Training rooms are quiet. Real deployments aren't.

Artūras Malašauskas is an AI Systems Integrator with 20+ years of production-grade web engineering experience. He has designed, shipped, and scaled enterprise Python/PHP systems for logistics, SaaS, and public-sector clients. For the past year, he has focused exclusively on AI integrations: deploying open-source LLMs, building generative media pipelines (image, audio, video), and engineering multi-agent workflows for real production environments. His standard: reproducibility, security, cost-efficient inference—no vaporware. He documents and evaluates emerging AI tooling, separating verified capabilities from marketing noise. Technical editor at: muza-ai.eu, ai-verslas.lt, ai-naujinos.lt Connect on LinkedIn