Data management and AI implementation are two sides of the same coin

Like all professional relationships, the interplay between enterprise AI and data requires care and caution—but a new survey suggests the two don’t always work well together.

The research, from Nasuni, a data platform that focuses on unstructured data and cloud storage, shows how mishandling the relationship can lead to security breaches, gaps in data quality, and misuse of AI technology.

Parag Pathak, Nasuni VP of product marketing, told IT Brew that the company surveyed 1,000 purchasing decision-makers from around the world to get a sense of how the technology is deployed and data is managed. The company’s State of Enterprise File Data 2026 report details how unstructured data is impacting enterprises as they determine the best approach to AI implementation.

“A year or two ago, the biggest thing around AI was the models—do I have the better model, do you have the better model,” Pathak said. “What this report is surfacing is that it’s no longer the model, it’s the upstream data.”

A staggering 94% of enterprises face problems from unstructured data, Nasuni found, but only 16% treat it as a top-three investment priority. Pathak sees it as a major missed opportunity.

“We are at an intersection where data is the key to drive advantage for your organization, because hardware costs are going up,” Pathak said. “A lot of our customers are dealing with it, where they think prices rise anywhere from 200% to over 1,000%. You can’t just outspend the data problem anymore.”

Change is coming. AI implementation is likely to change the fundamentals of data management, said David McJannet, co-founder of Dome Systems. He told IT Brew that AI solutions make data management easier in some ways—just point an LLM at a dataset and let it work, for example—while occasionally adding toil.

“I love the fact that I can now bring all this data over really, really quickly using AI, but at the same time sometimes it makes mistakes,” McJannet said. “That’s the trade-off you’re making, you’re going to either do it super quickly using the probabilistic model, or you can do it slower using the traditional models and traditional approaches that we’re all familiar with.”

How to handle it. For most IT pros, the success of AI implementation and data management will be tied together. Rima Safari, US data, analytics, and AI practice leader with PwC, told IT Brew that she sees it as a mutually beneficial relationship—one that IT teams will have to manage to find success and that organizations will need to invest in.

“AI’s value is going to compound when you invest in the right data quality, so data and AI will reinforce each other,” Safari said. “The best AI programs are not going to win without investing in the right data management capabilities upfront.”