Legacy data systems bottleneck AI agent adoption
45% of companies give AI agents access to enterprise data, laggards at 30%
New research from MIT Technology Review Insights, in partnership with Google Cloud, highlights how legacy data systems are crippling AI agent adoption. The report, based on a survey of 300 executives, found that AI agents currently access only 45% of enterprise data on average—dropping to 30% in 'data laggard' organizations. Meanwhile, 'data leaders' (those accessing >70% of data) report 100% trust in their agents' decisions, starkly contrasting with the 50% trust rate across all respondents.
The urgency is underscored by Gartner’s prediction that AI agents will handle 50% of business decisions by 2027. Legacy systems, even recently updated ones, fail to meet agentic AI’s demands for real-time, cross-enterprise data access. Two-thirds of laggards cite legacy constraints as barriers to scaling agents, while leaders report minimal issues. The top priorities for modernization? Improving access to structured/unstructured data and embedding business context into governance.
- AI agents access only 45% of enterprise data on average; 'data leaders' exceed 70% access
- Trust in agent decisions correlates with data readiness: 100% of leaders trust agents vs. 50% overall
- 69% of orgs plan to widely deploy agentic AI within 2 years, but legacy data systems block scalability
Why It Matters
AI agents can't deliver ROI without modernized data infrastructure—enterprises risk being left behind as agentic AI reshapes workflows.