A new report from MIT Technology Review Insights, sponsored by Google Cloud, finds that the gap between AI ambition and AI-ready data is widest in financial services, a warning sign for servicing and collection operations weighing where to deploy agentic AI.
The survey of 300 data and technology executives found that every respondent expects to use AI agents within two years, and 69% expect widespread use. Yet only about half trust that their agents make accurate and relevant decisions today. The report ties that trust gap directly to data readiness.
Access is the core constraint. On average, AI agents can reach just 45% of company data. A small cohort the report calls “data leaders” gives agents access to more than 70%, and all of them trust their agents’ decisions. Companies sharing 30% or less see far weaker results. Financial services providers rank last, exposing agents to only 33% of their data on average, followed by healthcare at 40%. The report attributes the shortfall in these regulated sectors partly to legal restrictions that put some data out of reach.
For servicing and collection shops, the findings map onto familiar pressures. Customer service is the most common agentic use case at 62%, the same contact-center and issue-resolution work that dominates collections. But agents acting on incomplete or poorly governed data raise the stakes under the Fair Debt Collection Practices Act and Fair Credit Reporting Act, where a wrong action can create liability rather than efficiency.
Legacy systems remain the bottleneck. Two-thirds of “data laggards” say old systems limit their ability to scale agents (66%) and slow decision speed (68%), versus just 8% of leaders. Half of all respondents say legacy data is hurting the ROI of their agent projects. Entrenched silos (50%), unstructured data (40%), limited real-time access (34%), and missing business context (32%) top the list of platform complaints.
