“BREAKING DEFENSE” By Sydney J. Freedberg Jr.
“Ordinary soldiers and civil servants have used the Army Data Platform to build thousands of data analytics. Now the service wants to scale up ADP’s success. The Army wants to bring in multiple vendors with an ever-wider range of easy-to-use tools to support an even larger population of lay users.”
_______________________________________________________________________________
“The service has a clear vision of what it aims to achieve with ADP 2.0, which should come online sometime in 2025. The goal is a vast flowering of decentralized, do-it-yourself analytics: Army soldiers and civilians, from program managers to supply sergeants, making sophisticated use of complex data to do their jobs, without having to get specialized training in data science.
That vision builds on the success of the current Vantage ADP, run by Palantir, and its 45,000 active users, many of them young soldiers or junior civilians, who’ve been creating an average of 700 new or updated analytic products every week for more than two years. ADP has “empowered [users] across all military and civilian specialties — e.g. logistics, human resources, artillery, operations research, cyber — and across primarily junior and mid-ranks,” the Army’s chief data and analytics officer, David Markowitz, told Breaking Defense in an interview. They’ve used it for everything from unit readiness updates to financial management and recruiting.
For ADP 2.0, the Army wants to bring in multiple vendors with an ever-wider range of easy-to-use tools to support an even larger population of lay users. But the service isn’t sure how to scale up ADP in a way that works both for its budget and contractors’ bottom lines.
The crucial question, Markowitz told Breaking Defense, is: “How do we scale this in a way that is efficient for the government and we reimburse industry for their efforts”?
As ADP 2.0 grows and evolves, “the government doesn’t want to buy the same thing multiple times,” Markowitz went on. “Yet, there’s some level of customization [by industry] that we want to pay them for.”
That’s why the Army’s official request for information, released Nov. 30, not only asks the usual questions about what kinds of products contractors could provide, but solicits industry’s detailed suggestions on how to structure the program. “What type of contract mechanism (e.g., OTA, multi-award task order ID/IQ) provides the most flexibility towards achieving the ADP 2.0 objectives?” the RFI asks. “What recommendations do you have for the Government in developing an effective intellectual property strategy? What data rights should the Government retain and require?”
“I’m absolutely interested in industry’s feedback” due the 19th, Markowitz told a packed conference — standing room only — hosted last Thursday by the Northern Virginia chapter of the industry group AFCEA. “Once we have that, we’re probably looking at another industry day” to thrash things out with interested companies, with a contract award expected later this year.
Official concept for Army Data Platform (ADP) 2.0 (Army graphic)
The Army isn’t buying a specialized tool for specialists in data science and software development, Markowitz emphasized to AFCEA, but “low-code/no-code” tools that a wide range of users can easily apply to help them do their jobs. “Data scientists are kind of a unique breed,” he said. “Our goal for our data platform is getting usage across the Army.”
“We want users to have easy access, but we also want it to scale to new projects,” Markowitz continued. The problem is, those are two different objectives that are traditionally met by two different kinds of contracts.
If you want simple software tools that large numbers of federal workers and military personnel can use to do their jobs day to day, like Microsoft Office, you usually buy “seats” — that is, the government pays per user. But if you want a sophisticated central database with an equally sophisticated toolkit, like the IPPS-A personnel management system, you usually pay per project.
Traditionally, “you buy licenses based upon user or seat, or you can buy licenses associated with a product or function,” Markowitz told Breaking Defense. But ADP 2.0 needs to be “somewhere in the middle,” he said. “That’s hard to put your head around, because our normal way of working with industry doesn’t think that way. … We really want feedback from industry on how to do that well.”
Another complication is that the Army wants to move from a single vendor (currently Palantir) to multiple vendors. That way users can mix and match the best features of different companies’ products, without being contractually “vendor-locked” by “one to rule them all,” Markowitz said.
For example, he explained, perhaps one company will have the best data storage and distribution, but another has better AI tools like large language models, and yet another superior visualization tools to turn all the complex data into easy-to-grasp graphics. What’s more, which company is on the cutting edge in any given area will keep changing on over time, and the Army wants to be able to grab the latest and greatest as it becomes available from whatever vendor.
“We really want to make sure the Army is positioned well for … getting best-in-breed and flexibility across industry and being able to evolve with industry as they evolve, which we think be best done with having multiple partners,” he told Breaking Defense.
So to make ADP 2.0 successful, “the area I think is the hardest actually is not technical, but it’s business,” Markowitz said. The Army knows what it wants, and it’s confident industry can deliver — but, he asked, “how do we buy this capability?”
ABOUT THE AUTHOR:
Sydney J. Freedberg Jr. has written for Breaking Defense since 2011 and served as deputy editor for the site’s first decade, covering technology, strategy, and policy with a particular focus on the US Army. He’s now a contributing editor focused on cyber, robotics, AI, and other critical technologies and policies that will shape the future of warfare. Sydney began covering defense at National Journal magazine in 1997 and holds degrees from Harvard, Cambridge, and Georgetown.
Comments