WASHINGTON: The Pentagon’s Joint Artificial Intelligence Center is deploying a combination of tech across DoD back office functions to dramatically increase the speed, volume, and accuracy of tasks that support warfighters. Known primarily for its AI work, JAIC’s projects also entail a range of tech that complements AI, including business process automation (BPA).

“[Back office functions] are the things that make the department run on a daily basis,” Bryan Lane, a JAIC deputy director who heads up its business process transformation portfolio, told me. “And this is truly for government writ large. These are all the things that happen behind the scenes that make government machinery move and move in an efficient way.”

Applying AI, machine learning, automation, and other tech to the business processes that underlie back office tasks is what Lane calls “workflow warfare.”

The goal? Improve DoD’s overall efficiency. “We’re only as fast as our slowest process,” Lane observed.

Given the military’s size and complexity, Lane has plenty of opportunities. DoD back office functions range from financial and health management to customer service.

Lane’s task includes looking across the entire DoD for opportunities to apply automation and other tech, including AI.

“I have the benefit of the broadest definition of artificial intelligence,” he said. “So it’s beyond deep learning, machine learning. This also includes rules-based systems and logical systems, expert system design, cognitive engineering — really teaching machines to think and operate the way humans do. Oftentimes, there is some type of human interaction. So, when I say AI, it includes automation and includes humans.”

One project Lane is heading up now is called HUNT X. The idea is to apply ML and automation to identify and match financial transactions, such as the correct accounts payable to accounts receivable for Army contractors. The volume is enormous.

The project, Lane said, has reduced the manual effort required from hours to minutes through automation.

Another project Lane highlighted is for Navy human resources. Dubbed RED-DA (Record of Emergency Data — Dependents Addition), it is used to automate Navy personnel files. When a sailor updates an address of record, marital status, or dependents on a benefits plan, it “kicks off a whole chain of events” that must subsequently happen within Navy’s HR IT systems, Lane notes.

This project identifies opportunities to automate manual tasks when someone updates a file. Lane calls this project a “purely robotic process automation play.” RPA involves automating manual tasks by having bots “observe” workers via graphical user interfaces (GUI) and then automate the tasks. This differs from automation methods that use scripts or internal application programming interfaces (APIs).

Through JAIC’s automation efforts, the tasks required after Navy personnel file updates have been reduced from 15 to 20 minutes to mere seconds. JAIC’s efforts have translated into a 50 percent increase in efficiency, Lane said.

Lane said the JAIC is in the process of scaling up both the Army and Navy projects. His team is also looking for opportunities and partnerships to implement automation across other DoD back office functions.

These projects tie in with JAIC’s Joint Common Foundation, which he describes as an “incubator for products and prototypes.” He also notes that DRAID — a multi-award contract to provide JAIC with data services — will involve “a ton of automation,” including automated data pipelines, automated data transformation, and the creation of synthetic data.

Going forward, Lane has high ambitions for what he and his team can do to make the DoD more efficient across back office support services. He is focused now on “enablers” for future AI, ML, and automation projects. This will entail identifying use cases, resources, and building nimble project teams to understand processes for smaller projects or products.

Lane says there are “amazing projects that are still kind of new and haven’t fully formed.” For instance, he’s looking to adopt AI to design active systems that work together, creating new data. The data created, in turn, form pieces of the overall workflow. Such a holistic view is necessary because, Lane says, “A lot of times we get focused on solving problems in a snapshot in time,” but it’s equally important to zoom out to the bigger picture.

Lane adds: “I have a mission to help people think about automation way beyond” some of the limited applications today.