AI agents—autonomous systems that can handle tasks end-to-end—are the latest AI buzz and are getting more attention from business leaders. But on the developer side, agentic AI still faces adoption hurdles.

• 87% of survey respondents are concerned about the accuracy of AI agents.
• 81% are concerned about the security and privacy of their data when using AI agents.
• And only one-third of developers actually use AI agents today, with half of developers having no plans to adopt them (citing accuracy, security, and privacy concerns).

TL;DR for leaders

Before rolling out AI agents, leaders should consider how to address the trust and safety barriers. This means ensuring strong data protections and validating outputs. Leaders should get buy-in from developers through pilot programs and open feedback, rather than mandating adoption from the top down.

And bear in mind that high-quality data (up-to-date, structured, well-organized) can make or break the success of any agentic AI project. Without good-quality data, AI agents can’t deliver accurate, reliable, or secure outputs. That’s one of the main reasons developers are hesitant to use them.

Want to build AI agents that actually work? Focus on your data first, not just the tools & technology. Bad data breaks even the best AI.