In a new Roadmap Report, 33 CIOs and CTOs share their challenges and insights on getting the data right, for the right use cases.
The importance and value of data has never been clearer. But given the rapid evolution of AI and related technologies and their infusion into business, the best ways to manage and use that data have rarely been more challenging or vital.
In three CIO Think Tank roundtables, held from June through August of 2025, IT leaders spanning diverse industries, from financial services to higher ed, health care, manufacturing, and retail, shared successes and speedbumps in managing the data that feeds AI initiatives across a wide swath of use cases.
[ Download the full CIO Think Tank Roadmap Report: AI-Native Networking ]
The roundtables were facilitated by John Gallant, enterprise consulting director at Foundry, along with Amy Machado from industry research firm IDC, CIO editorial leader Amy Bennett, and Sandy Ono, CMO of partner OpenText.
Challenged by data quality and sovereignty issues? Wrestling with staffing, skills, and education concerns? Searching for ways to reduce AI risks? Read on for ideas from the front lines.
Keeping up with AI: The governance rules are changing
AI use is advancing at warp speed, even as enterprise technology leaders grapple with foundational questions, starting with whether existing data governance policies still hold up.
Some CIO Think Tank participants emphasized that they do. “It’s just another opportunity to say all the existing policies and procedures still apply,” said one CIO. Others argued that, while the goals of governance still apply, the manner in which AI uses data necessitates fresh thinking and updated policies. “The business outcomes that you want might be the same, but if you were operating from a very different point of view and things worked differently, then you’re not going to succeed” with a business- as-usual approach said IDC’s Amy Machado.
Among the challenges facing AI builders is how to set the right business expectations. “The AI model life cycle is not like the application life cycle,” said Jamil Badrudeen, VP AI and Financial Engineering at financial giant State Street. Applications are put into production once the code works correctly; AI models “have to be tested and trained in every environment where they’ll be deployed,” he said. Data scientists and users aren’t aligned on when an AI model is “ready” to start production use.
Further, the workforce needs new skills; models need new infrastructure. Line-of-business leaders are excited about AI but remain blasé about investing in data cleanup. AI risks aren’t fully understood.
AI and tech leaders at the forefront of these efforts have tested numerous strategies for addressing the questions, challenges, and opportunities. But it’s still early days for AI, and what works at one company may not play at another. Context matters, and the think tank panelists — whose AI projects range from summarizing medical documents to evaluating environmental data from NASA’s Mars Rover — offered plenty of variability, uncertainty, and disagreement.
Even the easy cliché “It all starts with getting your data right” isn’t universally applicable. Some companies have undertaken a major data infrastructure and governance overhaul before wading into AI; others report cherry-picking their best data to use as their AI starting points, and still other CIOs recommend diving in and using the AI itself to surface problematic data.
Download the Roadmap Report to read more of the panelists’ key ideas, offering CIOs new possibilities for moving their own work forward at the pace of…well, at the pace of AI.
