Enterprises and the tech vendors serving them are in an arms race to stay at the forefront of an agentic AI market that's getting more sophisticated by the day. UiPath’s 2025 Agentic AI Report captures a market barreling toward deployment while quietly wrestling with the hard parts: governance, integration, and control.

End-to-end: Businesses are clearly captivated by AI agents' potential to improve oversight of complex workflows, enhance integration between disparate applications, and automate entire end-to-end processes. But ambition doesn't cancel complexity. UiPath reports that 56% of IT executives name security as their top concern, underscoring just how high the stakes are for end-to-end AI deployment.

Humans still required: Taming advanced AI takes more than tech. It demands firm controls and steady human oversight, a point that industry leaders keep coming back to. "I expect that RPA will orchestrate the agents. You need clear orchestration and governance," said Max Ioffe of Wesco Distribution. Wolters Kluwer’s Abhishek Mittal echoes the point: "We have generally found that human review is critical, and I don't see that going away."

Proceed with caution: Leaders are taking a measured approach, starting with internal, lower-risk projects to build expertise and tighten processes before scaling up. Upskilling programs like UiPath’s public sector initiative support that strategy, underscoring a key truth: agentic AI works best when skilled humans stay in the loop. The report in full can be read here.