AI Agents Hit Trusted Data Barrier as Leaders Exceed 70% Access
A 300-executive survey finds AI agents access 45% of company data on average, while data leaders exceed 70% and report greater trust, scale and success.
Summary
An August 12, 2026 MIT Technology Review Insights report, sponsored in partnership with Google Cloud and based on 300 data and technology executives, finds that inadequate infrastructure and data are blocking returns from agentic AI. Agents average access to 45% of company data, falling to 30% or less among data laggards, while data leaders provide access to more than 70% and report greater success. Only about half of surveyed organizations trust their agents’ decisions as accurate and relevant, compared with 100% of data leaders.
Agents require structured and unstructured data, business context, and real time access to operational systems covering supply chains, point of sale, and human resources. Among laggards, 66% say legacy systems constrain scaling and 68% say they impede decision speed, versus 8% of leaders reporting either problem. Every respondent plans to use agentic AI within two years, 69% widely, while Gartner predicts agents will augment or automate 50% of business decisions by 2027. Priorities are broader data access, governance enriched with business context, and, particularly among leaders, automated data management. The sponsored report was produced by human researchers, writers, editors, analysts, and illustrators, not MIT Technology Review’s editorial staff, with any AI tools limited to secondary processes that underwent human review.
Positives
- More than 70% data access distinguishes data leaders, which report greater success with AI agents than other surveyed organizations.
- 100% of data leaders trust their agents’ decisions, compared with only about half of all surveyed organizations.
- Just 8% of data leaders report either scaling or decision speed constraints from legacy systems.
- 69% of respondents expect to use agentic AI widely within two years, and every respondent plans some adoption.
- Structured and unstructured data access, contextual governance, and automated data management are emerging as clear modernization priorities.
Risks & concerns
- 45% average company data access leaves most enterprise information unavailable to AI agents across the surveyed organizations.
- 30% or less data access among laggards sharply limits the information available for agent decisions and actions.
- 66% of data laggards report legacy systems restricting agent scale, while 68% report slower decision making.
- Only about half of surveyed organizations trust that agent decisions are accurate and relevant.
- Legacy systems struggle to provide real time access to supply chain, point of sale, and human resources data.