A sponsored survey report of 300 data and technology executives, produced in partnership with Google Cloud, examines how legacy enterprise data systems limit the effectiveness of AI agents. It finds that on average AI agents only access 45% of company data, dropping to 30% or less among 'data laggards', while a group of 'data leaders' grants access to over 70% of data and reports 100% trust in agent decisions versus roughly half for others. Data leaders also report far fewer scaling and speed constraints (8% vs. 66-68% for laggards). Within two years all surveyed organizations plan to use agentic AI, with improving data access and governance context cited as top priorities.
Questions this post answers
What percentage of company data do AI agents typically have access to in enterprises?
On average, AI agents have access to only 45% of company data across surveyed organizations, according to a survey of 300 data and technology executives. Organizations classified as data laggards provide access to 30% or less, while a smaller group of data leaders grants access to over 70% of enterprise data, correlating with greater agent success. Enterprise teams weighing AI agent investments can follow data infrastructure trends on daily.dev.
Why don't enterprises trust their AI agents' decisions?
Trust in AI agent decisions correlates directly with data readiness and access. Only around half of surveyed organizations trust their agents' decisions to be accurate and relevant, while 100% of organizations classified as data leaders, those granting agents access to over 70% of enterprise data, report full trust in their agents' outputs. Teams building trust into autonomous systems can track data reliability practices on daily.dev.