When AI Becomes the Customer, Who Is Really in Control?

Written
Author
Guest Blog: Henning vorm Walde

Read our guest blog post from Henning vorm Walde, Executive Advisor. In October, Henning moderated our exclusive Payment Roundtable for partners, which focused on agentic commerce. He brought his experience as CEO of paydirekt and as a FinTech advisor to the discussion. Henning is also a mentor in our startup programs.
————————————————————————-
Agentic Commerce is changing more than how we shop and pay. It challenges how businesses create value, make decisions and organize themselves.
What happens when your next customer isn't human?
Imagine an AI agent finding the right product, comparing offers, negotiating the price and completing the payment. No browsing, no checkout, no human making the final click.
This is the promise of Agentic Commerce. But focusing on the technology misses the bigger story.
Agentic Commerce is not just changing how transactions happen. It is changing how businesses need to operate. When AI agents begin making purchasing decisions, companies must rethink customer relationships, payment processes and established ways of working. The winners may not be those with the most advanced AI, but those capable of adapting their business models and organizations.
This was my central takeaway from moderating TechQuartier's Payment Roundtable focusing on agentic commerce. Bringing together perspectives from technology, payments, compliance, risk management and law revealed something important: The technical possibilities are expanding rapidly, but turning them into sustainable business requires changes well beyond technology.

Pictured: Sebastian Maus from Oliver Wyman
From Human Intent to Machine Decisions
For decades, digital commerce has been built around human behavior. We search, compare, decide and pay. Technology makes the journey faster and more convenient, but the customer remains at the center of every transaction. Agentic Commerce challenges this model. Instead of navigating individual steps, customers can delegate tasks to AI agents that interpret preferences, evaluate alternatives and potentially execute transactions within predefined boundaries.
In his opening keynote, Sebastian Maus from Oliver Wyman explored the emerging landscape of Agentic Commerce, including developments and use cases across international markets. Perspectives from China and India illustrated how differently digital ecosystems can evolve. Sara Ghaffari from Visa brought the payment-network perspective into the discussion: What needs to happen when an AI agent moves from recommending a purchase to initiating one?

Pictured: Sara Ghaffari from Visa
The distinction matters. A recommendation influences a decision. An autonomous transaction executes it. And once software moves from advising customers to acting on their behalf, the implications extend across the entire value chain. Merchants need to rethink how they reach customers. Payment providers must enable new forms of authorization. Banks face new expectations around transaction processing and risk management.
The customer journey is no longer necessarily a journey made by the customer.
The Trust Paradox: More Autonomy, More Complexity
How does a merchant know whether an AI agent is authorized to purchase? How do financial institutions distinguish legitimate transactions from manipulation or fraud? Who is responsible when an agent makes the wrong decision? These aren't theoretical questions. They determine whether Agentic Commerce can scale beyond isolated experiments.

Pictured: Julia Quittek, co-founder of consalty, and Henning
The Roundtable explored these challenges from several perspectives. Julia Quittek from consalty examined the implications for anti-money laundering and financial crime prevention. Lars Schumann from Risk Ident contributed the fraud and risk-management perspective. Matthias Terlau from GÖRG explored the legal questions emerging as autonomous agents become involved in commercial transactions.
Together, their contributions highlighted a fundamental tension: Technology can delegate execution, but responsibility still needs a clearly defined home. Existing frameworks for authentication, fraud prevention and liability will need to accommodate transactions in which the person benefiting from a purchase is not necessarily the party initiating it.
The challenge is to make these processes work at machine speed without compromising security or creating unnecessary friction. This is where technological innovation meets the realities of established business systems.
The Real Disruption Is Organizational
Having spent much of my career leading payment organizations and complex transformations, I have seen how often promising technologies encounter barriers that have little to do with their technical capabilities. Agentic Commerce is likely to be no exception.
Consider a bank preparing to support agent-initiated payments. Technology may be ready to build the necessary interfaces. Product management sees a new customer proposition. Compliance needs clarity around authorization and liability. Risk management requires appropriate controls. Sales wants to move quickly. Everyone has a legitimate perspective. But who brings them together? Who sets the priorities? And who has the mandate to change processes that were designed for a world in which humans-initiated transactions?
This is where Agentic Commerce becomes a leadership and transformation challenge. From a systemic perspective, organizations are shaped by the interactions between people, structures, incentives and established decision-making patterns. Introducing autonomous agents into commerce affects these relationships. It can challenge existing boundaries between customer service, sales, payments and risk, redistribute responsibilities and expose organizational contradictions that were previously manageable.
The central question is no longer simply what AI agents can do. It is what organizations must change to make their capabilities commercially valuable and scalable.
This requires more than investing in new systems. It requires leaders to challenge existing assumptions, align competing priorities and create the organizational conditions for innovation.

Pictured: Henning with Eric Krausch, CEO at 1-CP
From Hype to Execution
So where should companies start? Certainly not by trying to automate everything at once. The contributions from Frank Meltke of contraco and Eric Krausch, including practical perspectives on implementation, helped connect the broader vision of Agentic Commerce with the realities of execution.
For executives and founders, the starting point should be a clearly defined business problem. Where would delegated decisions genuinely improve the customer experience? Which processes are sufficiently standardized to support greater autonomy? And where does human intervention remain essential?
A retailer might begin by testing AI-assisted replenishment for recurring purchases. A financial institution could explore mechanisms that allow agents to act within explicitly defined spending limits. Both are opportunities to learn how autonomous transactions change customer behavior, economics and operational requirements.
The point is not to demonstrate that an AI agent can complete a transaction. It is to establish whether that transaction creates measurable value. This requires cross-functional collaboration from the beginning, realistic performance metrics and the willingness to challenge assumptions as new evidence emerges.
Start with a hypothesis, not a technology. Test the value, not just the functionality. And build on what you learn rather than what you assume.
The Future Won't Be Built in Silos
One of the most valuable aspects of moderating the TechQuartier Roundtable was observing how the discussion evolved as different perspectives came together. What initially appears to be a technology question quickly becomes a payment question, then a regulatory question, a business-model question and ultimately a question of how organizations make decisions. The audience discussion reinforced the breadth of the opportunity, touching on topics such as stablecoins, B2B transactions and the role of regulation.
This is precisely where innovation ecosystems such as TechQuartier create value. Bringing together startups, established financial institutions, technology providers, legal specialists and regulators does more than encourage networking. It creates opportunities to challenge assumptions, identify interdependencies and translate emerging technologies into viable business models. Initiatives such as TechQuartier's Digital Finance Accelerator and its dialogue with regulatory stakeholders can help turn these conversations into practical progress.
The next generation of commerce will depend as much on new forms of collaboration and organizational adaptability as on technological innovation.
A New Question for Leaders
Agentic Commerce has the potential to fundamentally reshape customer relationships, payment processes and business models. But introducing AI agents into existing structures will not automatically unlock that potential.
The organizations that succeed will be those willing to rethink not only their technology, but also how decisions are made, how teams collaborate and where value is created.
The question is no longer whether AI agents will change commerce. It's whether our organizations are ready to change with them.
Are we building smarter transactions, or are we ready to build smarter organizations?
Henning vorm Walde is an Executive Advisor and Interim Executive specializing in strategy, payments and organizational transformation. As a former CEO in the payments industry, he combines executive leadership experience with a systemic approach to organizational development and change. He moderated TechQuartier's Payment Roundtable in Frankfurt on October 29, 2026.




