Updated
Updated · MIT Technology Review · Aug 12
AI Agents Access Only 45% of Enterprise Data, Legacy Systems Crimp ROI
Updated
Updated · MIT Technology Review · Aug 12

AI Agents Access Only 45% of Enterprise Data, Legacy Systems Crimp ROI

3 articles · Updated · MIT Technology Review · Aug 12

Summary

  • A survey of 300 data and technology executives found AI agents can access only 45% of enterprise data on average, leaving many organizations short of the foundation needed to deliver returns.
  • That gap is far worse for “data laggards,” where access falls to 30% or less, while “data leaders” give agents access to more than 70% and report stronger results.
  • Legacy systems are a key brake on scale and speed: 66% of laggards said old data platforms limit agent scaling and 68% said they slow decision-making, versus just 8% among leaders.
  • Trust tracks data readiness as well—only about half of surveyed organizations trust their agents’ decisions, compared with 100% of data leaders.
  • Pressure to fix data estates is rising because all respondents expect to use agentic AI within two years, with 69% planning broad deployment and Gartner projecting agents could shape 50% of business decisions by 2027.

Insights

If AI models are ready, why are legacy data swamps secretly sabotaging your company's autonomous future?
Are you risking catastrophic failure by handing autonomous AI agents the keys to your unstructured enterprise data?
With 2027 rapidly approaching, will wrapping legacy systems be enough to save data laggards from extinction?