The company Denodo has presented the results of its new study “The AI Trust Gap Report”, which reveals that 58% of companies in Spain consider that AI is only reliable when data is accessed in real time. In turn, 20% require data with a latency of less than one minute.
This shows that the immediacy of data is an essential condition for AI systems to generate accurate results that allow scaling towards more advanced and autonomous use cases based on quality data.
The Denodo report, which analyzes the trust gap between what organizations need to bring AI to production and what they are able to offer today in terms of data, also states that AI is an operational reality in Spanish organizations, with 68% already having AI initiatives in production and 32% planning new projects based on this technology and the intensive use of data.
However, as AI evolves from passive chatbots to agents capable of making independent decisions and triggering operational workflows, the importance of data accuracy has never been higher. Now, this requirement on data quality highlights the challenges that companies face in carrying out these types of data-driven initiatives.
The importance of data management for reliable AI
One of the main challenges identified has to do with the high dispersion of data. 44% of AI initiatives in Spain are based on between 250 and 549 different data sources, and 16% work with more than 1,000 data sources from multiple systems. This reality complicates access to the information necessary for AI projects, to the point that 32% of organizations consider it very difficult or difficult to connect and access data distributed between different environments.
Added to this is the difficulty of maintaining a coherent view of the data. 38% of companies encounter problems standardizing metadata that differs between systems, while 30% report difficulties clearly identifying which system acts as a reliable reference source for their data in AI initiatives.
“AI is rapidly moving from systems that simply answer questions to systems that take autonomous actions, and this transition completely changes the data requirement,” said Dominic Sartorio, vice president of Product Marketing at Denodo. “When an AI agent triggers a business outcome, there is no room for stale or uncontrolled data. To scale agentic AI with confidence, companies must move beyond static data silos and embrace a real, governed, context-relevant database.”
Governance, quality and compliance
The report notes that, beyond access to real-time data, trust in AI in Spain also depends on the ability of organizations to ensure strong governance, data quality and regulatory compliance. In this sense, 38% of AI teams identify regulatory compliance as their main data challenge, while another 38% recognize difficulties in ensuring the quality and cleanliness of the data used by AI. Added to this are the challenges in terms of data security, since 26% consider it very difficult or difficult to guarantee adequate access control over sensitive information.
Enterprises must move beyond static data silos and embrace a real, governed, context-relevant database
These results reinforce the idea that the trust gap in AI is structural in nature and does not respond to a limitation of the models, but to the way in which organizations manage and govern their data. Only by combining real-time data access with cross-functional governance, control, and quality will organizations be able to move from experimental AI initiatives to scaled, reliable use in data-driven decision-making and automation environments.
