While the concept of Agentic AI may be new to some, the tools themselves are not. Many employees interact with AI agents regularly, even if they don’t call them that. According to Wrike’s recent survey of 1,000 full-time knowledge workers, 64% say they use AI agents provided by their employer, and more than half of them use those tools daily. This suggests that Agentic AI may be more widespread—or more loosely defined—than previously understood.
Even the most advanced AI-powered tool can’t operate effectively if it relies on disjointed, inaccurate, or incomplete information. Many organizations still grapple with inconsistent data practices, creating barriers to effective AI deployment. Without access to high-quality, well-structured proprietary data, AI agents risk becoming more of a burden than a breakthrough.
Interest in Agentic AI is quickly turning into action, with nearly two-thirds of employees using the AI agents provided by their employers. Among those with access to an AI agent at work, 53% use these tools daily, and another 34% use them weekly, highlighting a clear appetite for automation when the right tools are in place.
The outlook is more skeptical among employees without access to Agentic AI. Only 29% say they’d use it daily, and 24% wouldn’t use it at all. But this isn’t necessarily indicating a lack of interest—it’s more likely a reflection of uncertainty.
Whether or not employees currently have access to AI agents, 87% of respondents report at least one concern about using them at work. Employees’ top three concerns are that an AI agent will:
- act on inaccurate or incomplete information (37%);
- introduce new data privacy concerns (31%); and
- fail to understand the context of their work (28%).
Only 12% of respondents cite concerns about an AI agent’s ability to work well with their team’s current tools and processes. Still, integration remains an issue.
To drive intelligent automation, organizations need access to well-organized data. This includes process or operational data, business rules and logic, customer data, user interaction data, performance metrics, and system and integration data. While 74% of employees say their company treats knowledge and data as valuable assets, only a fraction are managing it in a way that supports AI readiness.
When asked who’s primarily responsible for capturing knowledge and data at their organization, respondents were split: 24% pointed to project managers, 23% to team or department leads, and 21% to individual employees. Another 23% said it’s a shared responsibility. Without clear ownership over notetaking and feedback capture, documentation becomes inconsistent and unreliable, giving AI agents nothing solid to act on. Whether through dedicated roles or shared protocols, organizations must establish standardized methods of capturing knowledge to ensure AI tools have accurate, consistent context to work from.
