Agentic AI is moving from research to real-world enterprise use. Organizations are rapidly adopting agents to streamline workflows, boost productivity and drive innovation. "There have been several studies where there are discussions about every employee in an organization having up to four to five agents in the future working along with them to accelerate productivity," said Narayan Sundar, director of AI GTM at Palo Alto Networks.

However, this evolution comes with increased exposure to risk. AI agents are interacting with sensitive data and opening new threat vectors. "There are more threats and risk. Users are exposed to more data as well. That's one of the areas that needs to be protected," said Pritish Sinha, Senior Product Manager at Google Cloud.

It is essential to implement appropriate guardrails and safety measures from the earliest stages of experimenting with agentic AI to ensure secure deployment, Sundar said. "You don't want your company to be in the news for the wrong reasons."

In this interview with Information Security Media Group at RSAC Conference 2025, Sundar and Sinha also discussed: