Denodo study flags trust gap in agentic AI adoption
Denodo has published research identifying a trust gap in the adoption of agentic AI. The study surveyed 850 executives worldwide.
The findings suggest organisations are running into data problems as AI systems move beyond generating responses and begin taking actions tied to business processes. Some 66% of respondents said access to real-time data is non-negotiable if AI is to be considered trustworthy.
That requirement is colliding with increasingly complex data environments. The average enterprise AI initiative now draws on more than 400 data sources, while 20% of organisations manage more than 1,000 sources for their AI work.
Data discovery is another obstacle. Some 63% of executives said finding relevant data in the right business context remains a primary barrier to AI deployment, indicating that access alone is not enough if systems cannot determine which information suits a given task.
Security controls are also proving difficult to manage across fragmented estates. Some 67% of organisations struggle to maintain consistent security and access controls across systems, an issue that becomes more pressing when AI tools are expected to act autonomously rather than simply provide information.
Performance is another constraint. Nearly 60% of respondents reported difficulty optimising systems for the heavier workloads associated with large-scale AI use.
Data pressure
The report presents these issues as a data architecture problem rather than a model problem. Its central argument is that confidence in agentic AI depends on whether companies can supply live, governed and contextually relevant data at the point of use.
That distinction matters because agentic AI is increasingly seen as the next stage of corporate AI deployment. While earlier tools largely focused on answering questions, summarising documents or assisting staff with routine tasks, agentic systems are designed to make decisions and initiate operational workflows with less direct human intervention.
This shift changes the tolerance for error. If an AI system is linked to customer interactions, financial processes or internal operations, stale data or inconsistent permissions can create practical and compliance risks that do not arise in the same way with more limited chatbot-style applications.
The study points to a widening gap between ambition and readiness. Many companies want to move from experimental AI projects to systems that can support automated action at scale, but fragmented data estates continue to slow that transition.
Arlington Research conducted the study for Denodo. While the material released did not include a sector-by-sector breakdown, the global sample of 850 executives suggests concerns about data trustworthiness, governance and access are widespread rather than confined to a single market.
Shift in focus
The results also reflect a broader change in the corporate AI debate. Over the past two years, much of the discussion has centred on the quality of AI models and the pace of model development. This study instead shifts the focus to the infrastructure and controls surrounding data, suggesting these issues may now be the bigger obstacle to wider deployment.
That view is reinforced by the survey findings. Real-time access, contextual relevance, security consistency and system performance all point to operational issues in how data is stored, governed and delivered across an organisation.
Dominic Sartorio, Vice President of Product Marketing at Denodo, said the move towards autonomous systems raises the standard for data quality and governance.
"AI is rapidly shifting from systems that merely answer questions to systems that take autonomous action, and this transition changes the data requirement entirely. When an AI agent triggers a business outcome, there is zero room for stale or ungoverned data. To scale agentic AI with confidence, businesses must move beyond static data silos and adopt a foundation of live, governed, and contextually relevant information," Sartorio said.