AI agents are becoming a practical priority for business and technology leaders. The challenge is no longer only whether organizations believe in agentic AI. It is whether their data systems can support agents that need to decide, act and operate with trust.
A report based on a survey of 300 data and technology executives shows a clear pattern: companies with stronger data foundations are getting further with AI agents, while organizations held back by legacy data systems are seeing limits in access, speed and scale.
AI agents need more than stored information
Traditional enterprise data systems were built around reporting, search and analysis. Agentic AI raises the bar because the systems are expected to support action, not just answers.
That shift changes what data infrastructure must provide. AI agents need access to company data across structured and unstructured forms, and they need enough business context to interpret that data correctly.
They also need a path into operational systems. The source report gives examples such as supply chain, point-of-sale and human resources data. Without that connection, an agent may have information about the business but lack the real-time operational access needed to act on it.
This is where legacy data systems become a serious obstacle. Even systems that were updated just a few years ago can struggle when agents need broad data access, context and speed at the same time.
The access gap is still large
The survey found that AI currently has access to an average of 45% of company data across the organizations surveyed. That figure shows why many companies may be adopting AI agents while still falling short of the results they want.
The gap is wider among organizations described as data laggards. In those organizations, AI has access to 30% or less of company data. By contrast, a smaller group of data leaders gives AI access to over 70% of their data.
That difference matters because agentic AI depends on the quality and reach of the data environment around it. If agents cannot reach enough of the enterprise data estate, their decisions may be narrower, slower or less useful.
The report connects this divide with outcomes. Data leaders are seeing greater success with their agents than the rest, while organizations with more data limitations face more friction from legacy systems.
Trust follows data readiness
Trust is one of the central issues for AI agents because these systems are expected to make decisions and take action. According to the survey, only around half of organizations trust that their AI agents make decisions that are accurate and relevant.
The contrast with data leaders is sharp. The report says 100% of the data leaders trust their agents’ decisions. That does not mean trust appears automatically when an organization adopts agentic AI; it points to the role of the data foundation beneath the technology.
Reliable decisions require reliable inputs. Agents need data that is available, connected and grounded in business context. When those pieces are missing, confidence in autonomous action becomes harder to justify.
This also explains why data governance matters. The report identifies enhancing data and AI governance with business context as a high priority for organizations trying to scale AI agents.
Legacy systems slow scale and speed
The survey shows that legacy data constraints affect two of the most important promises of agentic AI: scale and faster decisions. Among data laggards, two-thirds say legacy data systems limit AI agent scaling (66%) and prevent agents from making decisions at speed (68%).
Data leaders report far less friction. Only 8% report either constraint. That suggests the main dividing line is not interest in AI agents, but whether the data environment allows those agents to operate effectively.
The pressure is increasing. Within two years, 100% of respondents plan to be using agentic AI, and 69% expect to use it widely. If data systems remain fragmented or difficult to access, wider adoption may simply expose the same bottlenecks at a larger scale.
For organizations, the message is direct: agentic AI programs need data modernization as part of the core work. The report points to improving access to structured and unstructured data for AI agents as the most important initiative for enabling scale among all respondents.
What data leaders are prioritizing
The survey does not frame data leadership as a single tool choice. Instead, it describes a set of capabilities that make AI agents easier to deploy, trust and expand.
Several priorities stand out:
- Broader data access: AI agents need to reach more enterprise data, including structured and unstructured sources.
- Operational connectivity: Agents need access to the systems where business activity actually happens.
- Business context: Data and AI governance must help agents interpret information in a way that fits the organization.
- Automation of data management: Data leaders are focusing heavily on this area as they prepare for agentic AI at scale.
The broader implication is that AI agent success depends on infrastructure choices that may sit outside the visible agent experience. Users may judge the agent by its actions, but those actions are shaped by the data, systems and governance available behind the scenes.
As more companies move toward wider agentic AI use, the organizations that remove data bottlenecks first are likely to have a clearer path to trusted autonomous action. The technology may be ready for broader use, but the enterprise data estate has to be ready too.