What if the scaling AI problem your agency is stuck on has already been solved by the agency next door?
For federal CIOs and chief AI officers, one of the most challenging aspects of scaling artificial intelligence is not about the technology. It is knowing what is already working at the federal agencies next door.
On Oct. 8, Leadership Connect hosted a webinar in partnership with Quantifind to close that knowledge gap. “The Agentic Government: Building Trusted AI for Mission-Critical Decisions” brought the CIOs of NIH and the Bureau of Industry and Security and the Air Force’s lead for AI test together with Adam Mulliken, Quantifind’s chief product officer.
Live attendees left the webinar with several actionable takeaways from Mulliken: a way to measure AI efficiency before costs rise, a plan to grow AI builders in-house, and an early look at new vendor due diligence rules.
Each takeaway starts with a big question agency leaders are already being asked, and the first one was about money: what happens when AI stops being cheap?
Measure Scaling AI Efficiency Before Costs Rise
CIOs, IT budget owners, and chief AI officers are pricing AI today at rates subsidized by venture capital, and those rates will not last long-term.
“The costs of this are not going to stay the same… I think the tokens will cost more in the future when they are not so subsidized in the way they are today by venture capitalists,” Mulliken explained. “We need to be able to measure efficiency and know what’s working and what’s not working.”
Mulliken’s point is that agencies do not need a separate cost-tracking system. The governance framework they are already building to manage risk can also measure efficiency.
Measuring efficiency only pays off if the agency has its own people building things worth measuring, and that’s where Mulliken turned next.
A Plan to Grow AI Builders In-House
“Even a company like Quantifind, and we consider ourselves an AI company and a fairly successful one, we do not stand a chance to recruit the premier AI scientists out of the frontier labs. So, we’ve got to develop the builders ourselves, and it is about upskilling everyone,” Mulliken said.
Federal payrolls cannot match private sector AI salaries. Mulliken is saying that an AI company can’t either. That makes the talent gap something every organization shares, not an individual failure on the agency’s part.
“And it is not just adoption of, ‘Hey, I want everyone trying the chatbot.’ We start there and really get everyone’s curiosity, but we want to encourage builders. And then once they have built something, we want to have some measures of effectiveness and efficiency and then celebrate those wins.”
An Early Look at New Vendor Due Diligence Rules
Mulliken said new rules taking effect at the end of the year will expand how many government vendors and suppliers need deep due diligence, which is “more than anyone is prepared to handle,” according to Mulliken.
For acquisition, procurement, supply-chain and vendor-risk leaders at Commerce, BIS, GSA, Treasury, DHS, and DOD, this is a workload increase arriving in a few months. Adding staff to the current process will not cover a 10x increase in vendor evaluations.
“Maybe you do QA on 10% of the work that’s maybe the most critical, or that’s how much budget you have for QA,” he stated. “That status quo of ‘90% gets no QA’ is maybe a bigger risk than taking AI and running it fully autonomously across 100% of the things that you do.”
Show Industry Partners How Scaling AI Really Gets Done
Whether the job is vetting vendors or anything else, Mulliken said the AI projects that succeed start with agencies showing industry partners how the work actually gets done.
“One of the things that we need as a partner is to really understand the process… We need someone who is going to come kind of open-minded, open-armed: here is how we do things today… Being able to map a solution to a specific problem is how we all succeed,” Mulliken emphasized.
That is a significant gap for many agencies: not a lack of AI tools, but a lack of visibility into what is already working down the hall. It is possible the agency next door has already worked through these problems, and partners like Quantifind already see those answers moving across the federal government and private sector.
Simply put, scaling AI faster means not solving every problem from scratch. It means solving them together.
- Watch the full webinar on demand to hear more actionable insights from Adam Mulliken and his fellow panelists, including:
- George Forbes, Chief Information Officer, Office of the Chief Information Officer, Bureau of Industry and Security (BIS) the U.S. Department of Commerce
- Dr. Jennie Khun, Chief Information Officer and Director, Office of Information Technology for the National Institutes of Health (NIH), U.S. Department of Health and Human Services (HHS)
- Maj. Brad Worley, Chief, Cyber, Digital and AI Programs, Test and Evaluation, U.S. Department of the Air Force
