Army’s New AI/LLM Tools Boost Productivity

U.S. Army leaders are forging ahead with the next steps regarding the new Army Enterprise Large Language Model (LLM) Workspace. This month, officials will strive to solidify the base of the platform and grow the number of use cases that the service can assist in achieving, according to Leonel Garciga, chief information officer at the Department of the Army. 

Additionally, the Army Enterprise LLM Workspace has only been available for a couple of months, but since its release, branch leaders have had the opportunity to receive feedback from public and private sector officials, Garciga mentioned during an exclusive interview with SIGNAL Media.

“I think the other piece is some initial feedback out to the larger community on what we learned from our ask of where do we have human-intensive processes,” Garciga explained. “And I think that’s going to spur a really good, very needy one-end list of where our priorities are, and where there’s going to be opportunity for industry to weigh in on some of our problems, and where we really need some artificial intelligence (AI)/machine learning to help give us a lift.”

Furthermore, Army leaders now have a better understanding of what has worked, what has not worked and what changes need to be implemented. With that in mind, this month, officials will also focus on analyzing and reviewing those lessons learned.

“I do think we’ll hear a little bit more about what we’ve learned and what guardrails we need and where we have found holes in our swing,” Garciga said. “I think you’re going to hear my big push. So, we’ve learned a bunch of things in this middle area on the platforms, needing industry’s help to really think through: What are the service level agreements? What language do we need to put in contracts? What the one-end difference models within the platform space are.”

“And really starting to get that feedback of where industry thinks the best approach is, whether it’s a platform I bought and you reach into my space or you have a platform and you deliver AI as a software as a service,” Garciga added. “I think there’s still a lot of conversation to happen there, because we’re seeing these different models pop up, so I think you’re going to hear about that in a big way.”

TechNet Augusta 2025 will take place later this month in Augusta, Georgia, at which Garciga said he plans to make additional comments about the Army Enterprise LLM Workspace project and these ongoing developments.

“I think what you’ll hear in Augusta from me is, here’s where we were at that point eight months ago, and here’s where we are today,” Garciga said. “And oh, my goodness, has the landscape changed.”

As aforementioned, these developments are emerging several months after Army leaders released the Army Enterprise LLM Workspace in May. The service is housed under the Army’s cArmy Cloud with the goal of helping soldiers and civilians alike by encouraging innovation, improving efficiency and providing a new avenue for more transparent communication, according to Army Public Affairs personnel. 

The Army Enterprise LLM Workspace consists of a plethora of features, including easy access because of the low barrier to entry, controlled unclassified information accreditation, a generative AI engine tailored to Army needs supplied by Ask Sage and a token-based subscription, according to Army Public Affairs personnel and Garciga, who illustrated an elaborate picture of these benefits in real-world situations. 

“The next kind of block is a block where we’re seeing a lot of momentum right now in a different way,” Garciga said. “The first one is, think ‘choose your own adventure.’ Users can go in, fully democratize out to the force low barrier of entry. The second piece is a piece that we’ve really seen take off over the last six to eight months, and really six months being the key piece where we have enterprise-level platforms where the vendors are coming in and bringing AI models to bear on problem sets really focused around the platform and the data in that platform.”

Furthermore, Software-as-a-Service (SaaS) providers are asking to use their LLMs and AI engines against the Army’s data set, which could lead to a smoother workflow and better user experience, according to the SaaS providers and Garciga.  

The Army Enterprise LLM Workspace has already proven to be helpful on and off the battlefield. One way the capability has been useful in combat is related to data sharing and soldiers’ ability to access data. Specifically, Army leaders began noticing that several command groups would ask similar questions about acquiring access to contract data. To address this inconvenience and make this data available to the individuals involved, Army leaders invited Army Contracting Command officials, the owners of the data, to prepare the data set and make it accessible to interested command groups, according to Garciga. The capability helped provide transparency and erased the need for anyone to ask repetitive questions.

“So, right now, a lot of the work that we’re seeing, which I think is really important work, is [an] area where there’s common enterprise data is having a responsible data storage for it, who owns that, so we’re not replicating the same data sets all over the Army,” Garciga said. “So, that gets expensive fast, and that’s one of the second-order effects that we saw. So, [what] we’ve been really working on, as we get those use cases, is to try to elaborate across commands to make sure that we know who the right person is for the data so they can have it prepared and then document where that is so folks know moving forward.”

The Army Enterprise LLM Workspace also assists in more off-the-battlefield and logistical use cases. For example, the Army Human Resources Command asked crews to update 30,000 position descriptions, a task that would have taken 45 people 30 days to accomplish if done the traditional way, according to Garciga.

Instead, by using AI capabilities within the Army Enterprise LLM Workspace, three people were able to complete the assignment in fewer than two weeks. 

Additionally, Army leaders are using the new AI capabilities to help them make decisions faster, especially when they are facing policy-related questions. For instance, with the implementation of these new tools, officials are taking an average of 35 days to triage a new Army regulation, according to Garciga. Prior to the existence of the Army Enterprise LLM Workspace and its tools, it could take officials up to 120 days to triage a new Army regulation, a change that Garciga called a “big deal.”

“It’s a big deal,” Garciga said. “Think about it, all of a sudden you have the ability to comb through all this data in a very quick way if you really know how to use the props, and they’ve received some good training to get that done, and it’s really given them that level of capacity. And from a warfighting perspective, it means we get more capability out faster. So, that’s a big thing.”