Rethinking IT for an Agentic Future

As organizations navigate the rise of generative AI, a fundamental shift is underway in how IT departments operate. Leading technologists are discovering that agentic AI—systems capable of making decisions, executing workflows, and interacting across multiple enterprise platforms—offers transformative potential beyond simple chatbots or copilots.

“It’s like my long-lost friend,” says Richard Mackey, CTO of healthcare company CCS, describing how AI agents handle tasks he finds tedious. He recently had an agent generate a presentation for the CEO that was ready to deliver with minimal modification—a process that would typically take him four hours.

This sentiment is echoed across industries as organizations seek ways to leverage agentic AI’s capabilities:

The Current State of Agent Adoption

• Forrester reports that 75% of leaders are adopting agentic AI, but only a small fraction have implemented meaningful production applications
• Even leading companies haven’t fully realized the promised value—the core question is whether this technology can truly move enterprise workloads
• Agentic systems require governing not just individual models but also the entire ecosystem in which they operate

Key Considerations for Implementation

As organizations build agentic capabilities, experts emphasize:

• Comprehensive evaluation: Continuously monitor how agents behave, what tools they use, and whether they’re achieving intended outcomes
• Data governance: Ensure data quality and security across all agent interactions
• Human oversight: Design systems with clear escalation paths for complex situations or when human judgment is required
• Architectural foundation: Build enterprise platforms from the outset to support a governed path from experimentation to production

At Palo Alto Networks, CIO Meerah Rajavel has seen firsthand the impact of agentic AI. Their internal Panda AI handles 82% of IT tickets and reduced operational costs by nearly 70%. The company is also using agents to automate RFP responses—cutting processing time from weeks to hours.

Rajavel notes that successful implementation requires a complete reimagining of workflows: “Before you build an agent, spend time understanding the data landscape and be prepared to completely rethink how work gets done.” This approach enables IT to move beyond reactive support toward proactive problem-solving and strategic innovation.