This article was based on the interview with Albert Castellana, Co-Founder & CEO at GenLayer by Greg Kihlström, AI and MarTech keynote speaker for The Agile Brand with Greg Kihlström podcast. Listen to the original episode here:
The race is on. Across the enterprise, marketing leaders are under immense pressure to deploy AI, automate processes, and unlock the efficiencies promised by an agentic workforce. We are moving with unprecedented speed, transitioning from AI as a clever assistant to AI as an autonomous actor—an agent empowered to negotiate, make offers, and resolve customer issues on behalf of our brands. The potential upside is enormous, promising a new frontier of personalized, scalable customer engagement. Yet, in our haste to innovate, we are collectively sidestepping a foundational question, one that keeps the most forward-thinking leaders up at night: when an autonomous agent makes a decision that costs the company millions, damages its reputation, or violates a customer’s trust, who is accountable?
This isn’t a hypothetical exercise for a distant future; it’s an active and growing risk. The “accountability gap” is the chasm between an agent’s autonomous action and our current frameworks for responsibility and governance. When a human employee makes a mistake, there is a clear path for recourse, escalation, and correction. When an AI agent hallucinates a product guarantee, denies a valid refund based on a flawed interpretation of data, or creates a biased marketing campaign, the lines of responsibility blur into oblivion. For marketing leaders, this is not a problem for the legal or IT department to solve in isolation. It is a core brand and customer experience challenge that threatens the very trust we have spent decades building. The brands that thrive in this new era will be those that address this accountability gap not as a compliance hurdle, but as a strategic imperative.
The New Paradigm: From Control to Delegation For decades, marketers have operated with a high degree of control over their tools. We input commands, and the software executes them predictably. Autonomous agents shatter this paradigm. We are no longer simply operators; we are delegators, entrusting significant aspects of brand interaction to systems whose decision-making processes can be opaque. This shift from direct control to high-stakes delegation is a source of both immense potential and significant risk. As Albert Castellana explains, this is a fundamentally new relationship between brands and their technology.
“You’re saying, okay, I want this agent to be representing my brand in front of the public, but I don’t even know how it’s thinking, what it’s doing, what promises it’s making, right? Did it hallucinate something, right? Did it, you know, deny something, let’s say, deny a refund that really should have been refunded, right? Can I get sued because of that? Am I responsible because of that, right?”
Castellana’s point cuts to the heart of the matter for any marketing leader. Every interaction an agent has with a customer is a reflection of the brand. An unconstrained agent can become a liability at machine speed, making unapproved commitments or alienating customers with illogical denials. The challenge is that these agents are, by design, non-deterministic. Their power lies in their ability to reason and adapt, but that very same quality makes their actions difficult to predict. We are deploying a workforce of billions of potential agents, as Castellana puts it, “transacting with each other at light speed.” Without a robust framework for oversight and accountability, we are not just scaling our marketing efforts; we are scaling our risk profile exponentially. The Leadership Mandate: Architecting the Guardrails As agents take over more baseline tasks, the role of human leaders within marketing organizations must evolve. The focus shifts from executing campaigns to designing and policing the systems in which agents operate. This requires a new set of skills centered on risk management, systems thinking, and ethical governance. The primary job of a marketing leader in the agentic economy will be to define the boundaries of acceptable agent behavior and ensure those boundaries are enforced.
“Humans don’t really have anymore like, you know, like base line job, but they will have their liability, their responsibility that things go well… I’m responsible for this thing to happen and I’ve got this team of agents, they’re doing this thing. And then the question really for me would be, okay, how do you create the railway, right? How do you create the borders, right? The constraints so that things work well.”
This concept of creating “the railway” is the critical strategic task for leaders today. It means moving beyond simply prompting an AI and instead architecting its entire operational environment. What are the non-negotiable brand safety rules? What data sources can the agent access? What is the explicit escalation path for a customer dispute that exceeds the agent’s authority? What constitutes a “high-quality” interaction, and how is that measured and audited? Answering these questions is no longer just good practice; it is the fundamental work of modern marketing leadership. We are becoming the architects and ethicists for our new digital workforce, and our primary responsibility is to build the constraints that prevent our powerful tools from driving the brand off a cliff. The Adversarial Future of Customer Experience We tend to think of customer experience as a relationship between a human customer and our brand, whether that interaction is with another human or a system. But the agentic economy introduces a third party: the customer’s agent. As individuals begin to deploy their own personal AI agents to manage their purchases, subscriptions, and service requests, the nature of CX will fundamentally change. These customer agents will not be swayed by clever branding or emotional appeals; they will be ruthlessly efficient, logic-driven entities programmed to secure the best possible outcome for their user. This sets the stage for a new, adversarial dynamic.
“I think ultimately, the customer’s agent will try to find a way to game your agent, right? It’s going to be trying to find a way because like, you’re a good brand, right? And your naive agent is just going to be there for agents to be attacking to say, okay, like, what’s the best way? I can search the whole internet to find for ways to basically like, you know, profit from this offer.”
This is a bracing but necessary realization. Our customer service and sales agents will be in a constant state of high-speed, automated negotiation with agents that are actively probing for loopholes, weaknesses in terms of service, and opportunities to exploit promotional offers. A “naive agent” deployed without robust, game-proof logic will be a liability, potentially costing millions in unintended discounts or refunds. This forces marketing leaders to think like security experts, stress-testing their systems not just for functionality but for exploitability. The future of CX isn’t just about being helpful and friendly; it’s about being robust, logically sound, and resilient in the face of adversarial automated systems. Building a Justice System for the Agentic Economy The sheer volume and velocity of agent-to-agent interactions will overwhelm our existing human-centric legal and dispute resolution systems. A minor contract dispute over a $1,000 transaction simply cannot be resolved efficiently by a legal system that costs hundreds of thousands of dollars and takes years to reach a conclusion. To operate at scale, the agentic economy needs its own native system of governance—a way to enforce contracts and resolve disputes at the speed of AI. This is where concepts like decentralized, trustless decision-making become less of a technological curiosity and more of a commercial necessity.
“…the legal system is going to be really like in stress…the human legal system just cannot scale to that level… I will put an escrow, and then the escrow will be smart enough to understand what actually if I fulfilled my promise, and then your agent will be able to say, yes, I want to enter into this contract… That’s what we are right now, right? The customers are feeling like they’re not in power, right?”
The idea is to create a digital jurisdiction where contracts are self-enforcing and disputes are adjudicated by a consensus of neutral AI models, rather than a single, potentially biased party. For a brand, this provides an automated, scalable “railway” for its agentic interactions. It creates a fair and transparent mechanism for accountability, where both the brand’s agent and the customer’s agent operate under a common set of enforceable rules. This isn’t about replacing human legal systems but augmenting them for the high-volume, low-value transactions that will define AI commerce. It’s about building the foundational trust layer that allows brands and consumers to engage confidently in a world run by agents.
The journey into the agentic economy is both exhilarating and perilous. The temptation to deploy first and ask questions later is strong, driven by competitive pressure and the promise of transformative results. However, true leadership lies in recognizing that with great technological power comes the profound responsibility of governance. The accountability gap is real, and it represents one of the most significant brand risks of the next decade. The conversations we have today about liability, constraints, and systemic trust will determine our readiness for a future that is arriving faster than any of us anticipated.
The task ahead is to shift our mindset from being masters of tools to being architects of ecosystems. It is about building the guardrails, defining the ethical boundaries, and implementing new systems of accountability that can operate at machine speed. While solutions like decentralized digital jurisdiction may feel abstract today, the problems they address are becoming more concrete with every new agent we deploy. The leaders who will win are not those who are merely fastest to adopt AI, but those who are the most thoughtful in governing it. They are the ones who are already working to ensure they never have to answer that impossible question: “Who do we fire for what the AI did?”
