About a year or so ago, Maj. Gen. Anthony Genatempo, program executive officer for cyber and networks for the Air Force and Space Force, drew a line in the contracting sand.
He told one of his contracting officers and their team to cut the number of days to award a contract from an estimated 14 months to 14 days.
“They all got very, very long faces on, and they said, ‘Well, what do you mean?’ Then there was a whole bunch of them doing their best Scotty impersonations from Star Trek, ‘You asked the impossible?’ ” Genatempo said during Federal News Network’s AI and Data Exchange.
“I said, ‘Well, that’s what I want to do.’ So I want to start off with asking what they think is impossible. The moment they apply even the most superficial capability of what NIPRGPT and our acquisition-oriented modules that we’ve attached to it can do, they’re going to go from 14 months to maybe down to eight or down to six months. They’re going to do that all on their own, without me even prodding or poking or having to raise my voice.
“And when they come in and they say, ‘We got it down to six months,’ I will pat them on the back, and we will celebrate with cake. We will really celebrate that achievement. And then the next acquisition that they come in and they give me a timeline of six months, I’m going to go, ‘How about 14 days?’ I’m going to keep pressing that technology onto them so they can truly grasp what I believe, and what Air Force Research Lab Chief Information Officer Alexis Bonnell believes is the never-ending capability. We are just scratching that surface.”
The Air Force launched NIPRGPT, an unclassified version of ChatGPT, in June 2024. It’s part of the Dark Saber software platform, an ecosystem where airmen and airwomen experiment, develop and deploy their own applications and capabilities. The Air Force wants the tool to be a testing ground that will allow the service to better understand practical applications of generative AI, run experiments, take note of problems and gather feedback.
Genatempo said his team is running parallel processes around the acquisition timeline to ensure they meet all the regulations and policy requirements.
“I am still the human at the end of the process that has the signature responsibility. We are not handing over responsibility to AI. We’re using it as a tool,” he said. “The analogy that I like to give is when our finance teams went from ledgers to spreadsheets, no one really had an issue with them doing that. It was just a tool to make them more proficient, and that’s exactly how we’re using AI in the acquisition workspace.”
Air Force sets straightforward goal for AI tools
This is one of several AI tools PEO Cyber and Networks has been developing over the last few years, specifically in partnership with AFRL.
Genatempo said he met Bonnell at a conference in 2022 and her speech sparked his interest in how to apply AI to improve the acquisition process.
“I’ve heard people talking about AI and what they want to do with AI. Quite honestly, it was always the we’re going to solve world hunger with AI. If you just gave us X millions of dollars and three years, we’re going to solve world hunger. Now I applaud everybody who is aiming that high. But I had some more immediate needs. It started really with boiling it down to, what can I do today with AI that can help me and help my folks with the business of Air Force acquisition?” he said.
“It really was a grassroots start there. I gathered a bunch of people,” Genatempo continued. “We then connected back with Alexis and her team to go through the dynamics of using AI. Quite honestly, the Kool-Aid that this nice lady makes I drink by the gallon. It has really propelled us forward in bringing this incredibly revolutionary tool into our workplace.”
Bonnell said working with Genatempo and his team helps answer the simple question of how the service can make the life of warfighters a little better every day.
“Just getting started and letting people find their ‘A-ha’ moment and also reduce their pain with these tools,” Bonnell said. “One of the things we talk a lot about is, how do we give people minutes back on mission? That, I think, is the soul of the collaboration, and it’s just been really key to not being afraid to dig in.”
Genatempo, who manages 2,700 personnel under his command and is responsible for a $21 billion portfolio that includes developing, producing, deploying and sustaining Air Force aerial networks, enterprise IT and cyber infrastructure, could have deployed AI tools in a number of different areas ranging from finance to logistics to the supply chain.
“I think that the biggest return on investment for Department of Defense acquisition of capabilities is applying this tool to our contracting timelines. If I can burn that down — and by burn that down, any contracting timeline could exist anywhere from eight months out to 18 months — the sooner we can get tools to the warfighters,” he said.
Creating an AI-powered market research tool for Air Force acquisition
One of the tools that falls under NIPRGPT is called AcqBot. AcqBot helps contracting officers cut the acquisition time from months to hours by applying automation and AI to the slowest and most time consuming parts of the acquisition process from requirements through contracting, Bonnell said.
AcqBot uses a co-collaboration model so that specific groups can customize and optimize certain elements to meet their needs as they come up.
“It actually takes the acquisition process and breaks it down and, in essence, provides elements of where AI really makes sense along that process,” she said.
Genatempo added AcqBot gives contracting officers access to a wide array of data on particular industries, helps them identify companies in the market and then start gathering more information on the sector before writing requirements.
“There are a lot of databases that are disaggregated and then there is people’s personal knowledge, which is like, ‘I met a person who met a person who met a person,’ ” he said. “So just that very initial phase of the acquisition process, I see that being accelerating to a much faster pace where we can go, ‘I now have a good handle as a senior leader on what our industrial base has to offer toward this capability.’ ”
Another example is MASC, which helps contracting officers improve the process of foreign disclosure information to United States allies.
“If anyone has ever been through that process, it is incredibly complex. It is incredibly hard, but it’s also critical to get right and to make sure that we have the types of relationships with our allies and the equipment and materials that we share,” Bonnell said. “Just seeing that type of tool be developed and see where it’s saving the process time, making sure we’re still respecting classification levels and things like that is really encouraging.”
Automating the ATO process at Air Force
A third tool still is in the early stages of development and testing uses AI to automate the authority to operate (ATO) process against the risk management framework.
Bonnell said the Air Force, as well as the Defense Department at large, has a lot of hope for this type of tool because the ATO process, while critical, can be time consuming and expensive and can delay innovations getting to warfighters.
“As an authorizing official, it’s understandable that as new technologies come out, for just one second, we can have a little bit of like hug your AO because it is not an easy job, especially because you’re really expected to almost be an expert in every technology and cyber,” Bonnell said.
“What has been really important, and interesting for us, in a lot of the co-learning we’re doing is really figuring out what is actually different or unique about the AI. A lot of people assume that it’s incredibly different to secure and incredibly different to use. Actually, what we find is a lot of the privacy law — a lot of the cybersecurity issues and elements — are actually quite similar. So then we try to be very thoughtful of just what are the small adjustments.”
She added the ATO tool is based on open source technology and vetted by PEO Cyber and Networks, the U.S. Cyber Command and other cyber experts across DoD.
The collaboration between PEO CN and AFRL on these three AI tools and others is paying dividends more broadly than just one PEO or even the Air Force at large, Genatempo said. They are sharing the tools and lessons learned broadly across the entire department and with civilian agencies.
“I know for a fact all of my fellow PEOs do not have the time or the opportunity to become AI experts and to share it with everybody else. They’re kind of busy doing their jobs. So that’s where I felt that our portfolio for the Air Force needed to be,” he said.
“The pipeline that we’ve built with AFRL is really providing that great foundation that people are starting to pop up and ask and look in our direction, and we’re just throwing an arm around them and saying, ‘Come on in. We had not thought of that. Or guess what? We have, and we’re about to save you six months of what you thought you needed to do on your own. We’ve already done it for you.’ ”
Discover more articles and videos now on our AI & Data Exchange event page.
Copyright
© 2025 Federal News Network. All rights reserved. This website is not intended for users located within the European Economic Area.
