FORT Belvoir, Va. –
The Defense Logistics Agency deputy chief financial officer is exploring the power of artificial intelligence to accelerate the agency’s path to a clean financial audit.
“Although we’re at the beginning stages of using AI in finance, we believe there’s a lot of potential. We’ve already had success with technology such as bots and robotics process automation. We plan to build on that with AI,” Shawn Lennon said.
AI can transform DLA’s progress by loading policies, data, process documents and more into a searchable large language model for easy, fast retrieval. The goal is to input data from DLA’s business systems and use AI to detect errors, generate insights, and propose solutions to improve data quality and financial reporting, Lennon added.
With AI, employees can approve an AI-proposed solution and execute the correction with robotic process automation.
“We’re also looking at using AI to reconcile DLA’s inventory in the Warehouse Management System with our financial records,” Lennon continued. “Today, we’ve got many people manually reviewing error transactions, trying to figure out what went wrong, why, and the potential solution. It’s too much for us to keep up with.”
The Defense Department must achieve a clean audit opinion by fiscal 2028, according to the 2024 National Defense Authorization Act. A draft of the 2025 NDAA directs the secretary of defense to report to Congress on the department’s use of AI to accomplish that.
Lennon volunteered to lead the DOD Financial Management Working Group to determine how DLA, the military services and other defense agencies can incorporate AI for audit purposes while protecting DOD data. The group is studying use cases from industry and other federal agencies and learning from experiences of DOD entities like the Defense Finance and Accounting Service, which has already solidified some governance on using AI for efforts like fraud detection.
The Government Accountability Office’s chief scientist recently briefed Lennon and participants in the Partnership for Public Service AI Federal Leadership Program on how they are using a large language model to search a century of audit reports to draw insights.
“DOD could do something similar. For example, if you wanted to know the history of the Joint Strike Fighter and every audit that’s been done, have it present summaries and recommendations, plus the status of implementation, AI can place all that at your fingertips,” Lennon said.
Although DOD agencies will likely use AI for audit in different ways, Lennon said the group helps everyone benefit from others’ progress and setbacks. Ideas of one agency may also appeal to another, added Navy Cdr. Johnathan Henson, an executive officer for DLA Finance.
“If there are solutions that could be helpful across different agencies, they could possibly pool resources to fund something together. And whatever one of us learns from a data-tagging or security perspective, we can all learn from,” he said.
DLA leaders also hope to learn from other federal agencies such as the State Department, which has spent the past year deploying AI on its internal network and will soon demonstrate its practices for DLA.
Using AI to store and sift through data is risky, however, especially when shared outside the agency because information can land in the wrong hands, Lennon said. GAO mitigated that risk by disconnecting its environment from the internet.
“That reduces some of your capability, though; so those are the trade-offs we need to decide on, even if the data we’re bringing together isn’t classified,” he said.
Since most AI large language models are web-based and not specifically for DOD use, DLA Finance has only limited experience so far. When it had temporary access to Microsoft Azure OpenAI, the team used it to load large documents such as the 1,500-page DOD Financial Management Regulation and conduct data summarization on specific themes.
The team is now working with DLA Information Operations to responsibly adopt AI tools and fund use cases.
“We need to determine how we’re going to prioritize what we do with AI and establish processes for approving and testing AI – all the stuff that comes along with any new technology,” Lennon said.
Whether DLA needs to create positions for employees who specialize in AI or train existing employees on skills such as prompt engineering are other considerations.
Though job security often enters employees’ minds with the mention of new technology, Henson said AI would allow employees to spend less time on manual tasks like managing spreadsheets and more time on meaningful work such as data analysis and decision-making.
Lennon added that employees should start thinking of the benefits of AI and seek training to better understand the possibilities.
“I want to inspire the workforce to raise their expectations for how we can use AI to help us be more efficient and effective in our mission, not just in the scope of finance and audit, but throughout the agency as we support the new Strategic Plan and objectives,” he said.
