Survey: Most companies limit AI agents to less than half of enterprise data

A new report based on a survey of 300 data and technology executives highlights the critical role of data infrastructure in the adoption of AI agents. The survey, conducted by MIT Technology Review Insights in partnership with Google Cloud, reveals that most organizations are not yet providing AI agents with sufficient access to enterprise data to realize their potential. On average, AI agents have access to only 45% of company data across surveyed organizations. This figure drops to 30% or less in organizations classified as 'data laggards,' which also report greater limitations due to legacy data systems.

The report identifies a subset of organizations—'data leaders'—that have largely overcome these constraints. These leaders provide AI agents with access to over 70% of company data and report 100% trust in the accuracy and relevance of their agents' decisions. By contrast, only about half of all surveyed organizations express trust in their AI agents' decisions, underscoring the link between data readiness and trust in agentic AI.

Legacy data systems are a major barrier to scaling AI agents. Two-thirds of data laggards report that these systems limit their ability to scale AI agents (66%) and prevent agents from making decisions at speed (68%). Among data leaders, who have largely modernized their data infrastructure, only 8% report similar constraints, indicating that overcoming legacy limitations is key to achieving scale and real-time decision-making.

The urgency to prepare data estates for AI agents is reflected in adoption timelines. All respondents plan to use agentic AI within the next two years, with 69% expecting widespread adoption. The report emphasizes that without addressing data system constraints, organizations risk failing to capture the efficiency and speed promised by agentic AI. Improving access to structured and unstructured data, as well as enhancing data and AI governance with business context, are top priorities for enabling scaling across all organizations.