Harnessing agentic AI: Governance, oversight, and practical steps for success

Agentic AI is the next evolution of artificial intelligence, moving beyond generating content to reasoning and executing tasks autonomously. For information professionals, this presents a chance to rethink workflows, strengthen governance, and collaborate with AI to drive greater organizational value.

Sue Trombley

December 15, 2025

I was recently joined by Swami Jayaraman, SVP and Enterprise CTO at Iron Mountain, for a chat about agentic AI for our December Education Series webinar. We explored how this next evolution of AI can transform information professionals’ abilities to manage data, boost productivity, and unlock new opportunities.

During this session, we walked through how organizations can navigate this shift and build the foundation to scale agentic AI safely and compliantly.

Moving beyond generative AI

Unlike generative AI, agentic AI doesn’t just respond to prompts—it acts, reasons, and executes complex tasks independently. AI agents can perform multi-step workflows and make decisions on your behalf, transforming AI from a passive assistant into a proactive digital worker.

The move toward agentic AI unlocks new opportunities for efficiency, productivity, and innovative ways of working across organizations. With repetitive tasks automated, information professionals can focus on higher-value responsibilities such as strategy, oversight, and exception handling.

We must start with non-destructive read-only tasks first, build trust and over a period of time slowly grant more authority to the AI agents.

Swami JayaramanSVP and Enterprise CTO, Iron Mountain

Safely scaling agentic AI with human oversight

Once foundational governance is in place, Swami stressed that scaling agentic AI depends on human oversight. People serve as the accountable authority by defining use cases, setting goals, approving high-stakes actions, and validating outcomes. This elevates employee roles from task execution to strategic direction and monitoring AI-driven workflows. By keeping humans in the loop, organizations ensure that AI decisions remain defensible and aligned with internal policies and regulations.

Scaling agentic AI safely isn’t an IT-only effort. In fact, it demands a coordinated approach across the organization. Centers of excellence (CoEs) bring together teams such as government affairs, legal, data privacy, information governance, and enterprise architecture to evaluate risks, interpret evolving regulations, and guide responsible adoption. By combining expertise across these functions, CoEs ensure AI initiatives align with enterprise standards while enabling faster, more confident deployment.

Human beings are the CEO of the agents we create.

Swami JayaramanSVP and Enterprise CTO, Iron Mountain

Implementing agentic AI in practice

Successfully deploying agentic AI begins with day-to-day operations and process integration. Organizations should start by mapping workflows, identifying repetitive or complex tasks, and determining where agents can meaningfully assist. Once these processes are understood, teams can begin integrating AI into existing operations in a way that complements established governance frameworks.

Pilot projects should follow a phased approach: begin with tightly scoped tasks, then gradually expand the agent’s authority as trust builds. This “crawl, walk, run” strategy, as Swami put it, ensures teams can validate effectiveness at each stage while maintaining control. With continuous monitoring, clear metrics, and feedback loops, teams are able to refine agentic AI performance and guide iterative improvements.

Human readiness is equally critical. Cross-functional collaboration and education programs ensure employees understand new workflows, when to intervene, and can collaborate with AI effectively. By equipping teams with knowledge and governance frameworks, organizations can unlock real value while staying compliant.

Always start with the people, always start with the processes first, and then you back your way into the technology bits.

Swami JayaramanSVP and Enterprise CTO, Iron Mountain

Moving forward with agentic AI

Agentic AI represents a significant shift in how organizations are able to manage information, streamline workflows, and unlock new opportunities. While the technology is powerful, its success depends on a strong foundation of governance, human oversight, and employee readiness. By starting with people and processes, defining clear rules and guardrails, and scaling thoughtfully, organizations can harness AI responsibly while driving measurable value.

Interested in hearing the full discussion? Visit Iron Mountain’s 2025 Education Series to watch the on-demand recording of “Agentic AI: What it is, how it works, and why it matters to information professionals.”