The merchant dilemma: How to stay relevant in the age of agentic commerce

While most conversation around agentic commerce revolves around consumer convenience and API connectivity, the real transformation lies in how merchants will maintain influence over the purchase decision.

Think back to when eCommerce shifted power from physical points of sale to digital experiences.

First, online shopping removed merchants from the counter, moving the checkout interaction from an in‑store cashier to a remote, software‑mediated flow. Then large marketplaces like Amazon, eBay, and Alibaba lowered the barrier to selling online by providing digital storefronts, traffic, payments, and logistics, so smaller sellers could skip building websites.  

Today, agentic commerce is going a step further. It removes what in the classic eCommerce world seemed impossible – the customer – delegating discovery, comparison, and purchase to autonomous systems – agents – that will treat brands as interchangeable suppliers.  

Those agents can decide, transact, and negotiate on behalf of consumers who may never even see a product page, logo, or marketing message.  

In practical terms, this means merchants risk losing control over pricing logic, bundling decisions and even which products are shown to consumers. When agents optimize for outcome rather than brand and AI executes entire purchase journeys, with humans stepping in only to approve – if at all – differentiation collapses.

Unless, of course, merchants actively participate in the decision layer.

The question for today’s merchants is not how to stay merely discoverable when agentic commerce goes mainstream – but how to preserve brand equity, reputation, and maintain control over the purchase flow.

Why APIs won't be enough: beyond 'discoverability'

In agentic commerce, the concept of a merchant's "brand" shifts from cosmetic aesthetics and human impressions to a set of machine-interpretable signals.

Today's public agents (ChatGPT, Claude, Perplexity, Gemini, and others) mostly crawl websites and extract information. In agentic commerce, "SEO" becomes "agent readability," while catalogs and checkout become machine-facing products that agents – not customers – are exposed to.  

One way to maintain relevance is to create connected, branded APIs that give agents standardized access to product catalogs, pricing, availability, and checkout flows.

For example, Stripe's ACP (Agentic Commerce Protocol), co-developed with OpenAI, defines a common language for agent-initiated commerce. This allows agents like ChatGPT to discover products, build carts, and complete checkouts directly with merchants without bespoke integrations. The public agent pulls your information, compares it across multiple merchants, and makes its own decisions on a customer's behalf.

But while API connectivity helps merchants become reachable by public agents, it does little to influence how decisions are made once access is granted.  

Merchant agents: optimizing for decision influence

APIs are designed to expose options, not negotiate outcomes. They allow merchants to set up integration protocols to make sure their products can be seen by consumers when making a purchase, much like marketplaces like Amazon allow anyone to start selling with merely uploading a product catalogue.  

But for merchants who want to preserve influence and outcome control over the purchase flow, connectivity alone isn't enough. They need a representative at the table, someone – or something – that can negotiate on their behalf, which a simple API connectivity cannot provide.

In agentic commerce, merchants must shift from optimizing for mere discoverability to optimizing for decision influence.

Unlike APIs, merchant-owned agents aren't just about exposing data. Much like having your own website or mobile application in an eCommerce world preserves brand integrity and narrative control, a bespoke agent allows merchants to protect margins and translate that hard-won brand equity into machine intelligibility.  

Merchant agents can represent pricing policies, fulfillment preferences, compliance constraints, and brand positioning in real time. They can adapt dynamically based on who they are negotiating with and under what conditions.  

With an agent under their belt, merchants get controlled agent-to-agent dialogue, ensuring the conversation about their products isn't entirely managed by ChatGPT or others.  

From eCommerce to Agentic Commerce

eCommerce is heading toward a marketplace ecosystem where public agents orchestrate purchases across merchants. This shift will happen faster than previous platform transitions – not necessarily because the technology is superior (although it is), but because the consumer behavior is already trained.

We have already spent years delegating to smart assistants that control our homes, banking apps that monitor spending and initiate transfers, and email and calendar agents that negotiate meetings and manage schedules with minimal input. We don't interact with those systems directly; we authorize them to act on our behalf. Commerce is simply the next domain where this delegation becomes default.

But there's a critical 6–12-month window where merchants who only focus on "being discoverable" will inadvertently undermine their own economics.

If merchants simply make their catalog available to public agents, they become single-SKU vending machines. A customer's agent finds one product match, executes one purchase, and moves on. Gone are the cross-sells, the upsells, the "complete the look" bundles, the basket-building logic that drives current margins. Revenue per transaction collapses.

The answer to this isn't to resist agent commerce entirely but for merchants to build their own agents before becoming discoverable to others. A merchant agent can understand the customer's full intent, apply the merchant's merchandising logic, and construct baskets in ways that maintains proper unit economics. Only after establishing this foundation should merchants integrate into broader marketplace discoverability.

Just as merchants who "won" eCommerce mastered personalized notifications and optimized checkout flows, the winners in agentic commerce will be those who build agents that negotiate, personalize, and preserve value on their behalf.  

Hilla Peled is SVP of Data Science at Nuvei, where she leads the company's AI and machine learning initiatives in payment processing and optimization.