Insight by Casepoint

Agencies look to AI, automation amid growth in digital records

Federal leaders see automation and AI as crucial to wrangling an ever increasing tide of digital records that's leading to backlogs in areas like FOIA.

Click here to watch the full discussion.

Federal agencies grappling with an “explosion” in unstructured data are turning to artificial intelligence and other process improvements to help manage the growth in digital records.

During a Federal News Network panel sponsored by Casepoint, officials representing  agencies ranging from the Army to the Federal Deposit Insurance Corporation described their efforts to modernize digital records management.

“What we’re seeing across government is that the volume of information is growing faster than people and traditional records management and disclosure processes can realistically keep up with,” Timothy Kootz, deputy assistant secretary for shared knowledge services for the Bureau of Administration at the State Department, said during the panel. “And on top of that, the public now expects information to be accessible, searchable and available quickly.”

Kootz pointed the State Department’s successful cable declassification pilot using AI.

“We’re already using AI tools that can do a declassification review faster and more consistently than a human team, and more importantly, these tools let us scale,” Kootz said. “We expect our workload and declassification specifically to cables to grow five-fold in the coming years, and without AI, we simply can’t keep up. So with these tools, we can meet the new growth in that demand without having to grow our staff in the same way.”

Embedded experts

Carrie McVicker, executive director for the Enterprise Services Agency at the Army, said her office sees similar challenges and opportunities, including with declassification and automation.

“One of the things we’ve done as well is embedded a records management expert within our declassification facility, because we found out while we’re doing some of this modernization, that sometimes we’re our own worst enemy when it comes to pulling records and reviewing them as we prepare for declassification,” McVicker said.

“Embedding a person in there and really strengthening that relationship between records management and declassification has really been key to some of the modernization work we’re doing, especially around architecture for declassification and preparing how we’re going to receive and send data through our declassification facility,” she continued.

Surfacing ‘critical knowledge’

While the FDIC doesn’t have to worry about declassification backlogs, the agency is still seeing a “massive explosion” of unstructured data, according to Richard Huffine, assistant director of enterprise information and records management at the FDIC. That includes emails, chat, video and other media managed in line with federal record retention schedules.

“We’ve implemented a capstone program that allows us to understand based on someone’s role, what their records may be at a broad termination. But we’re also looking at within that, how can we surface the most critical knowledge so that people are not working with 18 different versions of something, but working with the most trusted version that we need to keep long term, versus, files that need to be cycled through or not  valued as highly as that stuff that really tells the history and the decision making that’s done here at the FDIC.”

Data challenges across functions

The exponential growth in data holds implications across multiple mandates within an agency, including records management, the Freedom of Information Act, investigations and litigation, according to Kelly Swank, vice President of business development for government at Casepoint.

“With the implementation of AI throughout the government, the data volume is only going to increase probably five times more than what it is already today,” Swank said. “So we really are seeing people look at the full end to end life cycle challenge of data, because so many of these data sets across the different functions really play to each other. So it’s really important for them to have repeatable, automated and defensible workflows when they’re dealing with the data across the agency. We, of course, are integrating AI into our tools to really help streamline the process and automate a lot of functions for the agencies to speed things up so that the human review time is much smaller than it has been in the past.”

AI agents

J.D. Smith, deputy director of the Executive Services Directorate at Washington Headquarters Services within the Defense Department, said his team is looking to shift away from manual labor when it comes to many aspects of records management.

AI agents, he said, can “operationalize our policies, our file plans, our records disposition schedules.”

“We want to shift the human focus to more of an oversight and administration of the programs, not in the weeds, labor heavy [tasks], being the file clerk and things of the like,” Smith said.

Smith said his team is examining two related problems. The first is how to better manage existing records within the office of the secretary of defense, while also looking to improve management of records “at creation.”

That involves deploying “AI personas throughout the enterprise that are mapped to the records disposition schedules, that will operationalize our current manual process,” Smith said. “That’s performing the labor associated with identifying records, categorizing them, classifying them, filing them, executing preservation freezes, [requests for information], search, e-discovery, and also the embedding of missing metadata that is important, not only for our business need, but also our data leveraging needs to enable and create decision advantage for the secretary of war.”

Meanwhile, WHS is addressing how to improve records being create today.

“How do we manage records at origination, rather than after the fact?” Smith said. “So we’re looking to really manage using deep understanding, AI-enabled copilots set up similar to a file clerk, AI bot, AI enhanced RPA, that can analyze the content as it’s being created. So as you’re writing this file, you’ll get a pop up, ‘Hey, it looks like you’re making international legal agreement. If so, here’s the file numbers to select from. Please click this, and we’ll file it.’ We’ll embed the metadata for it, and we’ll manage that life cycle at origination.”

Copyright © 2026 Federal News Network. All rights reserved. This website is not intended for users located within the European Economic Area.

Related Stories