A new research report from BARC, co-sponsored by Ohalo, reveals a stark disconnect between enterprise confidence in AI scalability and actual data governance capabilities. While 79% of enterprises express confidence in extracting value from their files for AI without breaking governance, only 29% fully know where that data resides. This gap becomes critical as agentic AI raises the stakes for data mismanagement.
Confidence Outpaces Capability
The report, titled Harnessing Unstructured Data for AI Innovation, surveyed 225 enterprises across North America and Europe. It is the first in BARC's four-report AI series focusing on unstructured data—the files, emails, and documents that power AI. The findings indicate that enterprises believe they are ready for AI, but their own answers about data location, content, and governance tell a different story.
The Data Discovery Problem
Industry estimates suggest that roughly 80% of enterprise data is unstructured, holding institutional knowledge, sensitive information, and essential AI inputs. However, most enterprises cannot account for it. According to Kevin Petrie, VP of Research at BARC US, "Roughly two-thirds of AI adopters cannot effectively discover unstructured data, enforce governance policies on that data, or trace how they consume it. This hurts their ability to feed agentic AI the deep context it needs to take safe actions and generate business value."
The reality on the ground is often a stalled AI project. Teams greenlight initiatives, point them at file shares with years of institutional knowledge, but then cannot get past metadata issues. No one knows what is sensitive or fit for purpose. Work then shifts to manual discovery, initiatives break down, and ROI questions arise. This is the norm, not the exception: 70% of enterprises say less than half their data is even discoverable for AI.
Governance Held Together by Human Duct Tape
Kyle DuPont, CEO and Co-Founder of Ohalo, notes, "The governance holding all this together is human duct tape: 74% of enterprises still rely on manual effort or individual expertise. And only 29% fully know where their relevant data even resides. That's where AI initiatives stall, because when it comes to data, you can't govern what you can't find."
Recommendations for Bridging the Gap
BARC's recommendations include auditing and cataloging data before it enters production and closing governance gaps at the file level rather than at the perimeter. This approach aims to move enterprises from confidence to capability, ensuring that AI projects can scale without compromising governance or exposing the organization to risk.
