Sponsored by Tanium
Demonstrating how agentic AI can empower the federal workforce illustrates the art of what’s possible, according to Chris Kraft, Secret Service CIO.
May 27, 2026 2:08 pm
4 min read
The phrase “autonomous technology” seems nearly synonymous with AI, but it wasn’t always that way. The reality is that the government introduced autonomy into IT processes years ago. Fast forward to today: Developments in generative and agentic AI offer federal agencies a new era of transformational IT automation.
For all the furor around the use of AI in the federal government – some of it sensible, some of it a little more sensationalist – there are practical applications already being proven in use cases and prototypes. Readiness to scale more broadly? It’s not far off, according to Chris Kraft, the U.S. Secret Service’s CIO and chief AI officer.
“When you can demonstrate the capability and talk through the use case – even if it’s not yet fully perfected – I think that’s really helpful, especially for those who are not as technical, to see the art of the possible,” Kraft said. “I think one of the main challenges and opportunities in where we are with this AI technology is, can you be creative enough to leverage it within your mission space to deliver value that is going be either transformational or just incrementally better?”
Enhancing the human in the loop
As one example, agentic AI offers a clear and scalable path toward processing – and prioritizing for review as merited – the constantly growing volume of data that today weighs down agencies’ outdated and manual systems. The use of AI tools that scrutinize, organize and appropriately elevate data for review and response demonstrate how this new-era automation can empower the workforce and help people focus on essential areas for engagement, Kraft said.
The same goes for anomaly detection, where the placement of the human in the loop is based on predefined criteria for flagging items for individual review, he added. While this isn’t a new concept, agentic AI can take it a step further, taking some lower-level actions automatically while recommending for review and approval higher-level actions.
“I think that’s a great example of how you incorporate those types of decision points into the workflow so that a human has the opportunity to review and accept that course of direction,” Kraft said. “And I think as we look at scaling this out, whether it’s at Secret Service or other organizations, it’s about understanding what are you OK with and what can go ahead and be processed with the right controls and the right frameworks…then also having those defined criteria where a human does come in and make that decision and then the workflow can proceed.”
Real progress here hinges on testing and evaluation that keeps the technology on that clearly defined path – and also establishes a degree of comfort. Moreover, regular training and orientation reinforces and evolves understanding of capabilities, roles and decision-making across the workforce, Kraft noted.
“There should not be ambiguity when you’re presented with a decision…you need guardrails for the individuals that are going to be interacting with this capability,” he said. “To get the most value, it’s leaning forward on how we can leverage this within our workflows, to digest and process these large volumes of information and bring to the operators those critical insights that either you had a challenge getting to because of the volume of information, or because of the inability to perceive what these tools are able to perceive.”
Centering the mission-specific foundation
To be clear, it’s anything but a one-size-fits-all solution. When, where and how these AI tools are used depend on use cases, on the threat landscape, on the time to respond and on the decision-making structures in place, among other factors.
Kraft reiterated that a major key for success – especially for scaling beyond a prototype phase – involves solid testing and evaluation frameworks. These paradigms assess logic and provide transparent behind-the-scenes visibility into the agentic processes for determining next steps.
Building in robust governance, along with committing to continuous training and feedback from the workforce, are additional priorities that go hand in hand with testing and evaluation, Kraft said.
“What’s the technology available that could support your mission area? What does it look like and how can you leverage this? Because it’s about going back to the art of the possible and being creative in how you can apply this technology,” he said. “It’s making the workforce more efficient and delivering capabilities that they wouldn’t otherwise have the ability to deliver, either at the scale or the speed or the degree of precision. And I think it’s enhancing those capabilities that the individual team members can provide to the organization and empowering them in ways that really are transformative.”
Copyright
© 2026 Federal News Network. All rights reserved. This website is not intended for users located within the European Economic Area.