Global study: Poor data quality causes low trust in AI

66% says real-time data is a must for success with agent-based AI

A clear majority of companies worldwide say fragmented and outdated data is hindering the full adoption of agent-based AI, according to a recent survey from data management firm Denodo. The survey The AI Trust Gap Report shows that a critical trust challenge is hampering AI development today.

As AI evolves from passive chatbots to autonomous agents that can make their own decisions and carry out operational processes, the demands on data quality have increased significantly. At the same time, the study shows that technical barriers are a major challenge:

• The search for the right context: 63% of organizations state that it is a major challenge to find relevant data in the right business context.
• The need for real-time data: 66 % believes that data must be available in real time for AI to be sufficiently reliable.
• Security: 67 % states that they have difficulty creating uniform security and access controls across different systems, which is crucial for secure agent-based AI.
• Scalability and complexity: On average, companies today use over 400 data sources in AI initiatives, and 20 %s manage more than 1,000.
• Performance challenges: Nearly 60 %s state that it is difficult to optimize performance for the heavy workloads that large-scale AI requires.

The study also shows clear regional differences in how organizations define and manage trust in AI. In Europe, governance and compliance are particularly significant challenges. In France, 43 %s report that security and privacy are a significant barrier, compared to 20 %s globally. In Japan, trust in AI is strongly linked to stability and validation, while in the EU it is more closely linked to security, governance and compliance.

There are also clear differences between industries. The public sector lags behind in modern data management, especially in the Western world, where only 52 % use a lakehouse architecture, compared to 66 % in other industries.

AI is rapidly moving from a system that answers questions to a system that acts autonomously, and that is completely changing the requirements for data. When an AI agent drives a business process, there is no room for old or unstructured data. To scale agent-based AI with confidence, companies need to leave behind static data silos and instead build a foundation of real-time, governed, and contextual data, says Dominic Sartorio, Vice President of Product Marketing at Denodo.

The report, based on responses from 850 executives, shows that the trust gap is not about flaws in AI models, but about the underlying data architecture. To move from pilot projects to large-scale automation, organizations need to bridge the gap between fragmented data environments and the real-time data demands of agent-based systems.

About Denodo

Denodo is a global leader in data management, enabling reliable AI agents and applications. Denodo Platform, an award-winning logical data management solution, transforms enterprise data into trusted insights for AI, analytics, and self-service. Organizations worldwide use Denodo to deliver AI-ready and business-ready data significantly faster than traditional data lakehouse solutions, with up to 4X faster time-to-insight, 345 % ROI, and 10X better performance.