As organizations apply AI more broadly across the business, technology leaders are seeking paths to scale, doubling down on ROI, and developing the next generation of talent.
Executives shared their insights on these topics and many more at our May 2025 Metis Strategy Summit. Highlights are below. In the coming weeks, check out YouTube channel and the Technovation podcast for full panel recordings.
Leaders are actively measuring AI’s business impact. Hasmukh Ranjan, CIO at AMD, shared a bold productivity goal: “We will try to save 45 minutes of everyone’s time every day.” For IT, he tracks two KPIs: the ratio of employees to IT staff, and IT spending—arguing that the latter should decrease as AI investments yield results.
Motti Finkelstein, CIO at Intel, described the company’s focus on AI measurements: “We’re on a journey with AI in the area of time to market, revenue generation, as well as cost avoidance, and … savings. We track it very carefully.”
Shri Santhanam, EVP & GM of Software, Platforms and AI as well as Chief AI Officer at Experian, outlined how his organization’s four-pronged approach to AI implementation has helped them create value and get ahead of peers, with focus on building scalable platforms “to get out of POC land” and move toward enterprise-wide impact.
Agentic AI continues to be a hot topic for technology leaders, who increasingly are exploring how agents – bots that can complete tasks in addition to offering information — can improve their products and services.
A progression in AI applications emerged from the panel discussions. Santhanam described this evolution in four stages: First, knowledge bots and chatbots that provide general information. Second, copilots that offer personalized assistance while handling information securely. Third, agents that can complete tasks securely in a risk-managed way. And finally, agents that can handle multi-step, multi-task work across complex processes.
Santhanam noted trust as key to scaling agentic AI. Reflecting on learnings from developing Experian’s AI agent, Eva, he noted that “it is tremendously easy to make a PoC that looks good. It might even work with initial users and they will tell you it’s ok, but it can end up being garbage.” Their breakthrough came with a change in mindset, learning that accuracy and deep subject matter expertise was more useful than broad knowledge. “We don’t want a smart generalist, we want an accurate specialist,” he said. “This completely won the trust barrier.”
Chris Nardecchia, Chief Information and Digital Officer of Rockwell Automation, predicted that as agentic AI advances, organizations will need to create new roles to help it scale. He foresees an emerging role, “AI agent architect,” tasked specifically with addressing architecture, integration and access control.
Jaime Montemayor, Chief Digital & Technology Officer at General Mills, described how the company’s technical foundations are adapting to more advanced AI capabilities: “Tech foundations are evolving from just managing datasets to changing the way we ingest data,” he said. As agentic capabilities evolve, “there’s a need for a new semantic layer that we’re focusing on now.”
Technology leaders discussed various approaches to managing, organizing, and governing AI initiatives across their organizations. A clear theme emerged: the need to establish effective governance models that enable innovation while breaking down silos between technology and business teams.
Ryan Snyder, CIO of Thermo Fisher Scientific, noted the evolution in his company’s approach. “We went down the path immediately of centralization as an initial instinct that this could get out of control very quickly,” he explained. “We created a center of excellence to be an anchor point for the company, with the goal to then federalize and democratize it.” Finding the right balance between centralized governance and decentralized creativity was echoed by several executives.
Talal Butt, CIO at Generac, highlighted the need to break down organizational barriers: “We want to make sure we are not centralizing creativity; it needs to be freed up. We bring subject matter experts and people interested in experimentation together, breaking down walls between tech teams, SMEs, and partners.” This allows organizations to “thread the needle across different use cases.”
Allen Fazio, CIO at Houlihan Lokey, emphasized the need to create and communicate clear guidance on AI use. “Leadership needs to clarify the rules as quickly as possible,” he said. “The rules themselves help dictate how fast we can innovate.” This approach establishes guardrails while supporting controlled experimentation.
At Rockwell Automation, an AI community of support has become a resource for individuals and teams seeking to learn more about AI. “We believe in decentralization, but we want a place where people can come learn and experiment in a safe way, then teach the whole organization how to fish,” Nardecchia said.
Underpinning successful AI initiatives is a robust data strategy. Leaders emphasized that without quality data, even the most advanced AI tools would fail to deliver value.
Ramkumar Rayapureddy, CIO at Viatris, emphasized his team’s focus on creating accessible, high-quality data platforms. “Data is very important,” he said. “We have worked very hard to get data in one central platform.” In some ways, he noted, generative AI is helping this effort as the importance of data becomes even clearer..
This focus on data foundations was echoed by Hasmukh Ranjan of AMD, who noted that “data strategy is (the) center and core of our AI strategy, where we invest a lot of money, time, and people.”
As organizations move beyond proof-of-concepts to enterprise-wide implementations, Ampily Vijay, CDTO at CBRE IM, captured a sentiment shared by many leaders: “The big win is to scale AI – not just experiment with it.”
As AI transforms how work gets done, speakers emphasized the need to prepare their teams to use the technology effectively. Kristie Grinnell, CIO at TD SYNNEX, underscored the need for team members to have a mix of business knowledge, technical skills, and natural curiosity, and to treat AI like a powerful assistant rather than a replacement for human judgment.
“Who is the Robin to your Batman?” Grinnell asked, using the metaphor to illustrate how people should view their relationship with AI tools. “That’s how we should be thinking about AI and skills. Thinking about how you bring your Robin up to speed with you, to become Batman. It’s different because you have a sidekick, not a replacement.”
Madhu Ramamurthy, CIO at Zurich North America, emphasized the importance of subject matter experts: “Where I feel we can differentiate as a business is with SMEs,” he said. That includes a focus on product owners, “people who can define the why, what, and judiciously use the tools.”
Ramamurthy’s team focused on two critical areas. They prioritized building models that can be explained and understood and implemented an incremental rollout strategy to build trust. “The workforce starts with skepticism, (so) it’s important to release AI in an incremental fashion,” he said. This methodical approach has allowed them to extend AI into core business areas, including underwriting.
Sesh Tirumala, CIO at Western Digital, described the company’s systematic approach to talent development: “We go through our existing organization and do skills assessment. How do we go from here to there?” He noted the need for an expansive approach to talent development, with education just one piece of the puzzle. “It’s education, exposure, and experience.”
Motti Finkelstein, CIO at Intel, highlighted the advantage of pairing AI tools with early-career talent. “We want junior financial analysts to be in the room,” he emphasized, noting how this combination creates powerful learning opportunities. “We see tremendous uptick with junior developers – they absorb a lot more with generative AI helping them code.”
Leaders emphasized the need for pragmatic approaches to managing change during AI implementation. Neal Sample, CIO at Walgreens Boots Alliance, introduced the concept of “five minutes of ‘flawsome,’” an opportunity for teams to share lessons learned in addition to their successes and showing that failure is part of the process. “Behind failure was learning and opportunities,” he said. “You don’t call it failure; you call it a pivot.”
Snyder emphasized the need for executives to lead by example when it comes to AI adoption. “Role model leadership is incredibly important,” he stated. “How are you using these tools? Are you just delegating? How much are you walking the walk?” He stressed the critical importance of “ensuring that leadership is staying active in usage” rather than simply mandating adoption for others.
Darlene Taylor, CIO at Superior Industries, shared a leadership philosophy that resonated with other leaders navigating transformations: “Listen, drive, and care.” She explained, “It all starts with listening – to the board, strategic leadership team, functional groups, and getting down to the operator level. They’re the ones closest to your customer.”
Taylor emphasized the importance of driving both the business and technology forward while keeping stakeholders’ needs in focus: “We have to support the strategic leadership team and board with data, data, data and have playbooks ready.” This data-driven approach to change management aligns with what other leaders described as essential for successful technology implementations.
Jeffrey Katzenberg, Founding Partner at WndrCo and former CEO of DreamWorks Animation, drew parallels between identifying promising entrepreneurs and storytellers: “Great venture capitalists are constantly battling sensibilities of dreaming and at the same time being skeptical.”
On AI’s impact, Katzenberg said, “It’s not that AI is going to replace people, it’s going to replace people who don’t use it, understand it, or master it. Human inspiration and poetry won’t be replaced, but there will be tools that allow people to become much more efficient.”
Varun Mohan, Co-Founder & CEO of Windsurf (formerly Codeium), an AI coding assistant platform, shared their approach to innovation: “We believe AI is getting so much better every year that every six months our product should feel silly […] At any point, we need to be cannibalizing our product.”
As AI becomes more integrated into business processes, building trust remains paramount. Ather Williams III, Head of Strategy, Digital, Innovation, & Enterprise Payments at Wells Fargo, explained: “The core question around AI is trust. With AI moving quickly, how do you get people to trust the technology? It’s not just about sticking AI into your process; it’s about rethinking your process.”
Cindy Hoots, CIO at AstraZeneca, emphasized the importance of transparency with stakeholders in integrating AI into products and services: “Having a trust relationship is the most important thing,” she said. The company was among the early movers to publish a statement for the ethical use of AI in the pharmaceutical industry, highlighting how the company planned to use it.”
This emphasis on transparent, responsible AI implementation was echoed by other leaders throughout the Summit. As AI capabilities become more powerful and autonomous, establishing clear governance frameworks, ethical guidelines, and communication strategies will be increasingly critical to maintaining stakeholder confidence and driving successful adoption.
