At a moment when artificial intelligence (AI) is dominating headlines and investment cycles, government leaders face a more urgent and complex question: how to turn that momentum into real mission outcomes.
For Nikhil Krishnan, chief technology officer (CTO) of C3 AI, the answer starts with a fundamental shift in mindset – away from chasing technology trends and toward delivering operational value.
“We’re at a unique moment in time with AI,” Krishnan said, pointing to both rapid advances in model capabilities and the infrastructure buildout supporting them. “All industries are in scope for AI transformation, and government is no exception.”
That transformation, he emphasized, is already reshaping everything from back-office functions to national security operations.
“Everything is touched – and can be shaped by – AI,” he said.
From pilots to production
While many federal agencies have experimented with AI, fewer have successfully scaled those efforts into production. Krishnan drew a sharp distinction between pilots that stall and those that deliver sustained impact.
“One big factor is thinking about operational value first or mission value first, versus technology first,” he said.
“Pilots that think in terms of, ‘Hey, this is the end state that I’m trying to get to from an operational standpoint or a mission standpoint,’ and then design the technology around that, those are the ones that are set up for success from day one,” Krishnan explained.
C3 AI’s own work reflects that philosophy, with deployments that are deeply embedded in mission-critical government operations.
One example is the company’s work with the Defense Logistics Agency (DLA), where AI is being applied to the challenge of contested logistics.
“The challenge there is – what happens if in a contested situation, if a logistics node is destroyed, or a route is blocked, or materiel is destroyed,” Krishnan said.
Through AI-driven decision support, the company has been helping DLA to simulate disruptions and respond in real time, helping to “ensure that our warfighters have the right materiel, at the right place, in the right time.”
C3 AI is also working with the U.S. Marine Corps to modernize workforce planning. Through the use of AI, the Marine Corps can get a forward-looking sense of its staffing needs, recruitment pipeline, and training processes.
“Previously, this was a very manual task,” Krishnan said, noting that the shift to AI-enabled decision-making “really changes the game for how they manage their staffing levels.”
“We’ve always had a production deployment mindset to AI,” he added, focused on “actually deploying [AI] deeply in workflows to make change happen on the ground.”
Cutting through the hype
Despite growing AI adoption, Krishnan warned that several misconceptions continue to slow progress. One of the biggest misconceptions, he said, is the belief that AI models alone can drive transformation.
“It’s not just the AI model … that’s five to 10% of the solution,” he said. “There’s the other 90% that needs to be built in or needs to be purchased or extended or configured for specific use cases.”
He also pushed back on narratives around widespread workforce displacement, instead describing a shift toward human-AI collaboration and redesigned workflows.
“It’s going to be a reimagining of the workforce,” Krishnan said. “You are going to need humans together with AI to process most of the workloads that we’re going to see. There are going to be a lot of things that operate autonomously, but you’re going to need humans in new roles managing and operating these AI systems.”
Start with mission, not models
Looking ahead, Krishnan urged federal leaders to focus their AI strategies around a small number of high-impact use cases tied directly to mission outcomes.
“The way I would recommend federal leaders think about maximizing AI impact is to think about it from a mission and a use case standpoint,” he said.
That means identifying where AI can move the needle most significantly, then working backward to align data, technology, and processes.
Equally important is speed, the CTO stressed. With technology evolving rapidly, waiting for the perfect conditions can mean falling behind.
“The technology is also evolving fast, and the best way to get in the swim lane with the technology is actually to leap in, to jump in, and to pick those few use cases and start working on those,” Krishnan said.
Choosing the right partner
As agencies scale AI, partner selection is becoming a critical success factor. Krishnan stressed the importance of long-term alignment to avoid vendor lock-in.
“I’d be very careful about partner or vendor selection in those major use cases … I would think about who can be a partner for the long term,” he said.
C3 AI differentiates itself, he explained, through a platform-centric approach that allows agencies to build and expand multiple use cases on a unified foundation.
The company emphasizes interoperability and data ownership, ensuring that government customers retain control over their systems and can evolve them over time.
“Government really wants modular, open architecture systems, and we have done the utmost to comply. Every part of our stack is fully open, fully extensible, fully interoperable with other systems,” Krishnan said. “We’re not owning any of the government’s data.”
He said that C3 AI is “trying to also avoid vendor lock-in,” noting that customers can “pull out the data in the system” if they choose to move on.
Anchoring transformation in outcomes
Ultimately, Krishnan sees successful AI adoption as less about technology selection and more about disciplined execution around mission priorities.
For government leaders navigating competing demands, his advice is to focus on the use cases that matter most.
“I would encourage leaders to think about their specific mission profile, where the value pools actually are, and what the use cases are that would unlock those pools,” he said. “Then work backwards from there to the data, change management, technology selections, etc.”