Agentic Commerce: “Europe is Shaping Trusted Adoption, China Demonstrates Speed and Scale”

Agentic commerce is reshaping online shopping: instead of clicking through product pages, AI agents act on consumers’ behalf — reordering household goods, booking trips, managing subscriptions. But once machines start spending money, new questions arise around identity, authorization and human control.

In this interview, Connie Cheng, Vice President Innovation Asia Pacific at Mastercard, explains why trust and transparent consent are becoming the new infrastructure of commerce, how Europe and China are taking different paths, and what Mastercard’s recent entry into stablecoins through the BVNK acquisition means for the future of payments. She also addresses the risks — pointing to cases like OpenClaw, where AI agents have crossed boundaries humans never intended.

Trending Topics: Why is Mastercard currently focusing on agentic commerce? What significant value does agentic AI offer in e-commerce?

Connie Cheng: Agentic commerce represents a major evolution in digital commerce, where AI agents carry out tasks on people’s behalf within clear rules and permissions, rather than simply responding to commands. This shifts online shopping from manual interaction to delegated execution and makes trust, security and governance essential from the start.

Agentic AI adds value by reducing friction and complexity in e‑commerce, helping consumers make better decisions more efficiently while opening new opportunities for merchants. When designed responsibly, it can simplify decision‑making without removing human control, allowing commerce to scale in a way that remains transparent, secure and trustworthy.

Can you provide some specific use cases for AI-driven commerce?

In agentic commerce, trusted AI agents can search, compare and act within parameters defined by the consumer. Practical use cases include automatically re‑ordering household goods, booking travel when prices drop, or managing subscriptions and renewals. The value lies in convenience, smarter decision‑making and reduced friction for consumers, while helping merchants increase relevance and conversion.

As AI agents begin initiating transactions, it becomes essential that payments are built on a foundation of trust and transparency. Systems must know who is acting, on whose behalf, and with what authorization. Mastercard’s role is to help enable verified identity, secure consent and safe payments, so AI‑driven commerce can operate responsibly at global scale.

What data will actually be used to make purchasing decisions in the future?

Future purchasing decisions will be based on a combination of consumer‑defined preferences and real‑time context. This includes factors such as budget limits, preferred brands or merchants, delivery expectations and loyalty benefits, combined with live information on price, availability and delivery conditions, all used with clear user consent.

Equally important are trusted signals around identity, intent and authorization. For AI‑driven commerce to function safely, systems must understand who an AI agent represents and what it is allowed to do. Mastercard’s role is to ensure this data is used securely and responsibly, so AI can assist with decisions while people remain in control.

How does Europe compare internationally on this issue? How far along is China, for example?

Regions are progressing at different speeds. Europe is moving forward with AI‑driven commerce in a measured way, with a strong emphasis on privacy, consumer protection and accountability. Innovation is happening within clear regulatory frameworks designed to build trust.

China, by contrast, is further along in large‑scale deployment, supported by highly integrated digital ecosystems where commerce, payments and identity are already closely connected. These approaches are complementary: Europe is shaping trusted adoption, while China demonstrates speed and scale. Mastercard works globally to ensure agentic commerce can operate across markets while maintaining consistent standards for security and consumer control.

Do you think advertising for e-commerce products will work differently in the future? Does the agent need to become the recipient of the message?

Advertising will evolve as AI agents play a larger role in commerce, but it will not simply shift from humans to machines. In agentic commerce, marketing becomes less about persuasion and more about relevance and clarity, ensuring products are accurately represented when AI agents evaluate options.

The consumer remains the ultimate decision‑maker, setting goals and boundaries, while agents act within those parameters. Mastercard’s role is to help ensure this new model operates with transparency and accountability, so recommendations and transactions reflect genuine consumer intent rather than opaque automation.

Walmart recently stated that conversion rates for purchases made directly within ChatGPT were three times lower than those made directly on the website. What if ChatGPT and similar platforms don’t turn out to be as great shopping platforms as some people think?

Early conversion rates reflect a phase of experimentation. AI‑native shopping experiences are still new, and it takes time for trust, habits and seamless payment flows to develop.

Over time, what will matter most is that AI‑initiated transactions are grounded in clear consumer intent, trust and secure payment mechanisms, regardless of where the interaction begins. Agentic commerce will succeed not because of any single platform, but because the underlying infrastructure allows AI‑driven purchases to be reliable, transparent and widely accepted.

Do you see a risk that agentic commerce will shift even more market power toward the big platforms, away from smaller retailers? If AI agents go shopping, will online stores become less important?

Agentic commerce does not automatically favor large platforms. Because AI agents compare offers based on consumer‑defined criteria such as price, availability and service, they can actually increase visibility for smaller retailers that are competitive and trustworthy.

Online stores and brand experiences remain important, but the path to purchase changes, with AI handling part of the evaluation process. The opportunity for merchants of all sizes is to ensure their offers are transparent, reliable and easy for both people and AI agents to assess.

Mastercard recently made a major entry into the stablecoin business through its acquisition of BVNK. Will stablecoins become the currencies of commerce agents?

Stablecoins are one of several payment mechanisms being explored in the context of agentic commerce, but it is too early to say that any single form of money will dominate. What matters most is that commerce agents can transact using trusted, widely accepted and well‑governed payment rails, regardless of the underlying instrument.

There are repeated cases, such as with OpenClaw and other AI agents, where these agents actually break established boundaries and then do things that humans never intended. When it comes to money, commerce, and consumption, isn’t it very dangerous, in your view, to open this field up to agents?

These concerns are valid and underline why agentic commerce cannot be approached lightly. AI agents should never operate without clear boundaries, explicit authorization and accountability—especially when money is involved.

This is why agentic commerce must be built on strong safeguards from the outset: verified identity, clear consumer intent, secure transactions and the ability to understand and audit what an agent has done and why. Without these foundations, autonomy does not scale. With them, AI can support commerce in a way that remains safe, predictable and firmly under human control.