AI Agents See Only 45% of Company Data—and It's Costing Enterprises

The agentic AI era is colliding with a stubborn reality: most companies simply don’t have the data plumbing to support it. A new survey of 300 data and technology executives, conducted by MIT Technology Review Insights in partnership with Google Cloud, found that AI agents across surveyed organizations can only access an average of 45% of company data. That’s not a foundation for autonomous action—it’s a bottleneck wearing a trench coat.

The gap between data leaders and laggards is stark. Companies categorized as ‘data laggards’ give their AI access to 30% or less of their information, while a select group of leaders clears the 70% mark. The payoff isn’t subtle. Every single data leader—100%—reported trusting the decisions their AI agents make, compared to roughly half of organizations overall. Two-thirds of laggards say legacy systems actively prevent them from scaling agents (66%) and slow down decision-making speed (68%). Among leaders, just 8% report either constraint.

What’s driving the divide? Legacy data systems, even ones refreshed only a few years ago, were built for a world where AI answered questions rather than took actions. Agentic AI needs frictionless access to operational systems—supply chain, point-of-sale, HR—along with the business context that makes those data points meaningful. When agents can’t see the full picture, they can’t act on it. The report identifies the top priority for scaling as improving access to both structured and unstructured data, followed closely by enhancing data governance with proper business context.

The clock is ticking. Gartner predicts AI agents will augment or automate half of all business decisions by 2027, yet the survey shows 100% of respondents plan to be using agentic AI within two years, with 69% expecting widespread deployment. That’s a collision course between ambition and infrastructure. The organizations treating data readiness as a prerequisite rather than an afterthought are the ones positioned to actually deliver on the ROI promises that have so far remained largely theoretical. The rest are building agents on sand.