This article was based on the interview with From PegaWorld: enGen’s Richard Rutkowski on moving agentic AI from theoretical to practical by Greg Kihlström, AI and MarTech keynote speaker for The Agile Brand with Greg Kihlström podcast. Listen to the original episode here:
For the past several years, the conversation around AI in marketing has been dominated by two primary functions: prediction and generation. We’ve become adept at using machine learning to forecast customer churn, predict the next best action, and identify high-value audience segments. More recently, generative AI has captured the collective imagination, promising to streamline content creation, from ad copy to email subject lines. These are powerful, valuable applications. But they still position AI as a very sophisticated assistant—an analyst, a writer, a co-pilot. It provides insights and assets, but a human is still firmly in the driver’s seat, interpreting the data and orchestrating the subsequent actions.
What happens when that model evolves? When AI transitions from assistant to agent, from providing inputs to orchestrating outputs? This is the shift to “agentic AI,” a concept that is rapidly moving from the theoretical to the practical. It represents a move toward autonomous systems that don’t just suggest a course of action but actively execute complex, multi-step processes across an organization’s technology stack. While this may sound like a distant future, the groundwork is being laid today, and surprisingly, some of the most compelling blueprints are emerging from industries where the stakes are highest. In the complex, highly-regulated world of healthcare, leaders like Richard Rutkowski of enGen are building the architectural and strategic frameworks required to make agentic AI a reality, offering lessons that marketing leaders would be wise to study.
The Strategic Imperative: It’s About Scale, Not Science Fiction
For many, the idea of autonomous AI agents conjures images of futuristic, all-knowing systems. The reality, as is often the case, is far more pragmatic. The primary driver for adopting an agentic model isn’t the pursuit of a technological novelty; it’s a direct response to overwhelming operational pressure. In any large enterprise, the sheer volume of signals, workflows, and required actions has surpassed the capacity of human teams to manage effectively. This is the core challenge that agentic AI is designed to solve.
As Richard Rutkowski explains, the forces driving this shift in healthcare have direct parallels in the marketing world.
“The driving need to shift, though, to move to an agentic world, I think, is really scale… you’ve got aging populations with chronic illness on the rise. We’ve got clinician burnout to levels we haven’t seen before. So there’s a work shortage. Uh, regulatory policy… means more faster, and again, clinicians are burning out… if you want to scale, you can’t go with traditional care management, which is fragmented, disconnected.”
Swap out a few key terms, and the story becomes instantly familiar to any marketing leader. Instead of aging populations, consider the exponentially growing number of customer touchpoints and data signals. In place of clinician burnout, think of marketing teams stretched thin by the demand for hyper-personalization across dozens of channels. And the pressure of regulatory policy is a burden every global marketing organization understands intimately. The fundamental problem is the same: the old way of doing things—fragmented, disconnected, and reliant on manual orchestration—simply does not scale. Agentic AI is not a solution in search of a problem; it is the logical evolution required to manage the complexity we have already created.
You Can’t Bolt an Agent onto a Broken Chassis
The allure of a new AI tool is strong. The market is flooded with vendors promising to “unleash the power of AI” with a simple software integration. However, a truly agentic system cannot be simply bolted onto a rickety, siloed marketing stack. The prerequisite for an AI that can take meaningful, autonomous action is a unified and orchestrated operational foundation—what Rutkowski’s team refers to as an “architectural chassis.” This underlying platform is what allows an agent to perceive an event in one system and trigger a cascade of coordinated actions across many others.
Rutkowski provides a powerful example from the healthcare journey that illustrates the importance of this connected architecture.
“A discharge event is a data event that comes from an EMR, letting you know that somebody was in the hospital… What we like to do is several things. An agent can look at that, find it, determine the criticality of it… Then it would update the authorization in one system. It would start a case for care and schedule it with a skilled case manager in another system, and it would send a nudge to the member via digital means… that’s all orchestrated through this architecture.”
This is the holy grail for customer journey orchestration. Imagine a similar scenario in marketing: A high-value customer visits a specific product page but doesn’t convert (the “data event”). An AI agent could instantly assess this customer’s lifetime value, check their recent engagement history, and determine their eligibility for a specific offer. It could then simultaneously update the customer profile in the CDP, trigger a personalized follow-up email with a unique promotion, add the user to a retargeting audience in the ad platform, and create a task for a sales development representative in the CRM—all autonomously and in real-time. This level of sophisticated, cross-channel orchestration is impossible when your data, workflow, and intelligence layers are not built on a common chassis. The hard work isn’t just about training the AI model; it’s about building the plumbing that gives the agent its hands and feet.
Building Trust is a Process, Not a Proclamation
Even with the right strategy and architecture, deploying autonomous systems in a high-stakes environment is a delicate matter. The fear of AI running amok, making costly errors, or alienating customers is real and valid. Getting buy-in from the operational teams who will rely on these systems requires a thoughtful, incremental approach centered on building trust. It’s not about replacing human expertise but augmenting it, and proving the system’s reliability every step of the way.
The key, according to Rutkowski, is to start with a “human in the loop” model and gradually increase autonomy as confidence grows. This pragmatic approach demystifies the AI and demonstrates its value by tackling low-risk, high-friction tasks first.
“With Agentic AI, you can now take all the clinical data you have… and compare it to the medical policy and render a decision. But it can be done in parallel to what the medical director’s doing till you feel comfortable enough that, yeah, they’re aligned all the time… So, you can shift it left, if you will, from that parallel processing to putting it right in front of the medical director… to saying, ‘You know what? We’re just gonna let that go because it makes sense.’”
This “shift left” methodology is a perfect playbook for marketing leaders. Instead of flipping a switch and allowing an AI to autonomously manage a budget or launch a campaign, start in parallel. Have the AI recommend audience segments, but have a marketing manager approve them. Let it generate campaign performance summaries with insights, but have an analyst verify them. Over time, as the AI’s decisions consistently align with your experts’ judgment on routine tasks, you can confidently automate those approvals. This frees your most valuable talent from the drudgery of administrative oversight and allows them to focus on strategy, creativity, and the complex edge cases where their expertise truly matters. It reframes AI from a threat to a force multiplier, supplementing your team’s ability to do what they do best.
The journey toward agentic AI is less a technological sprint and more an organizational marathon. It begins not with a data science project, but with a strategic recognition that current operational models are unsustainable at scale. From there, it requires a deep, architectural commitment to unifying data, workflows, and intelligence onto a coherent platform. This is the unglamorous, foundational work that separates organizations that can merely talk about AI from those that can actually deploy it to drive meaningful, autonomous action across the enterprise.
While the specifics of enGen’s work are rooted in healthcare, the principles are universal. Marketing leaders stand at a similar crossroads, facing their own pressures of scale, complexity, and burnout. The future of a truly agile brand lies not just in smarter insights or faster content, but in building an operational nervous system capable of acting on intelligence with speed and precision. The path forward requires a pragmatic, human-centric approach to adoption that builds trust and empowers experts. The question for leaders, then, is not whether agentic AI is coming, but whether the chassis of their marketing organization is ready for the journey.
