The Agentic Commerce Paradox: Why Merchants Should Adopt UCP — On Their Terms

Tim Williams — CEO and Cofounder — AstraSync AI

At NRF 2026, Google announced the Universal Commerce Protocol with Sundar Pichai on stage and a coalition of 20+ partners endorsing the standard. Shopify, Etsy, Wayfair, Target, Walmart helped develop it. Visa, Mastercard, Stripe, PayPal, Adyen signed on. The message was clear: agentic commerce has arrived, and the infrastructure is ready.

The protocol itself is genuinely useful. UCP creates a standardised way for AI agents to discover products, negotiate checkout parameters, and complete purchases across merchant platforms. It works with existing payment rails. It’s compatible with other emerging protocols like AP2 and A2A. It solves real interoperability problems that would otherwise fragment the market.

Merchants should adopt it.

They should also understand what they’re adopting, and negotiate terms that protect their interests. Because the same week Google announced UCP, Amazon was pursuing litigation against Perplexity for building shopping features that surfaced Amazon’s product data. The parallel is instructive.

What Amazon Understood

When Perplexity launched shopping recommendations that pulled from Amazon’s catalogue, Amazon didn’t negotiate. They sued. This wasn’t because Perplexity acted with malicious intent. Perplexity was solving a genuine problem for users who wanted to discover and compare products through conversational interfaces. They were building value for their customers.

Amazon understood what was being intermediated: customer intent, browsing behaviour, cross-sell opportunity, the relationship itself. They had the resources to protect it through litigation.

Most merchants don’t have that leverage. When a platform with billions of monthly active users offers a protocol for agentic commerce, the negotiating dynamic looks different. Refuse participation and become invisible to AI-mediated discovery. Accept and gain discoverability. The terms, however, deserve scrutiny.

I explored this dynamic in a previous piece on the Amazon-Perplexity lawsuit, examining what happens when platform economics collide with the agent economy. The tension hasn’t resolved. It’s scaled.

The Coalition Question

The list of UCP endorsers reads like a who’s who of payments and commerce infrastructure. This coalition isn’t evidence that UCP serves merchant interests. It’s evidence that the parties endorsing it benefit from merchant participation.

Payment networks process transaction volume regardless of who owns the customer relationship. AI platforms gain structured access to product catalogues and checkout capabilities. Payment processors see the same upside: volume flows through their rails whether the merchant or the platform controls discovery.

The question isn’t whether these parties are acting in bad faith. They’re not. Google built UCP to serve users who increasingly expect AI interfaces to handle commerce end-to-end. Payment networks are positioning infrastructure for a market that’s clearly emerging. Everyone is acting in their interest.

The question is whether merchant interests are aligned, and what merchants should demand as a condition of participation.

What’s Actually at Stake

Richard Crone, CEO of Crone Consulting, published an analysis of UCP that quantifies the merchant risk with uncomfortable precision. When checkout happens inside an AI interface rather than on a merchant’s storefront, merchants keep Merchant of Record status. Google becomes System of Record for buyer intent data.

Crone values that intent data at $267 to $736 per monthly active user across Google’s 4 billion MAU base. Cross-sell and order-more-items revenue, which represents 33–76% of incremental sales for many retailers, shifts to whoever controls the agent conversation. Retail media revenue, often the second-highest margin line for large retailers, faces pressure when product discovery happens off-platform.

These aren’t speculative projections. They’re the economics that follow from moving the point of purchase from merchant storefronts to AI interfaces.

The operational impact is equally concrete. When a consumer checks out through Gemini instead of visiting a merchant’s website, the transaction arrives with a payment token and shipping address. What doesn’t arrive: device fingerprint, IP geolocation, session behaviour, browsing patterns. No cookie consent banner loads because the consumer never visits the site. The signals merchants rely on for fraud decisioning simply aren’t present.

The liability, however, doesn’t shift. Merchants retain full responsibility for chargebacks on transactions they have less data to evaluate. I covered this data gap in detail during Cyber Week, when the first wave of agentic transactions revealed the mismatch between liability and information.

The Expedia Lesson

The hotel industry learned a version of this lesson with online travel agencies. Properties that refused to list on Expedia didn’t maintain their independence. They lost visibility and declined. Consumers moved to platforms, and properties that weren’t listed became invisible.

Yet properties that listed without strategy became commoditised: interchangeable rooms sorted by price, with the platform owning the customer relationship and the cross-sell. The hotel became logistics. Warehousing with beds.

The hotels that thrived took a different approach. They participated on OTAs for reach and discoverability. They simultaneously invested in loyalty programmes, direct booking incentives, and owned channels. They used platforms strategically while retaining enough customer data to compete on experience rather than price alone.

The restaurants that navigated Uber Eats successfully followed the same playbook. Participate for reach. Build direct relationships in parallel. Never cede the customer relationship entirely.

Agentic commerce presents the same strategic choice. Merchants who refuse UCP risk invisibility as consumers increasingly discover products through AI interfaces. Merchants who adopt without protections risk becoming invisible logistics providers: warehousing and shipping infrastructure for transactions they don’t control and customers they don’t know.

What Data Parity Means

Merchants evaluating UCP adoption should demand data parity as a condition of participation. This means the same information available for on-site checkout should be available for agent-mediated transactions.

Bank-grade agent identity verification. When an agent transacts on a merchant’s platform, the merchant should know who built that agent, who deployed it, and whether the developing and deploying organisations have been verified to the same standard applied to payment counterparties. Agent registration with cryptographic tokens is a start. Full KYC-equivalent verification of developers and owners is the standard merchants should require.

Contextual metadata passthrough. The prompt context, behavioural signals, and session information that would normally be available on-site should pass through the verification layer. Merchants shouldn’t have to accept transactions blind simply because they originated from an AI interface.

Audit trails for disputes. When a transaction is disputed, merchants need documentation of what instruction was given, what constraints were in effect, and what the agent’s trust score was at transaction time. Without this, merchants are defending chargebacks with less evidence than they’d have for a standard e-commerce transaction.

Equivalent fraud signals. Device fingerprinting, IP geolocation, and behavioural analysis are standard for on-site checkout. Agent-mediated transactions should provide equivalent signals through alternative mechanisms, or the liability allocation should reflect the information asymmetry.

The UCP specification explicitly leaves agent identity verification to “other layers.” That’s appropriate architectural scoping. Commerce protocols shouldn’t try to solve identity. The question is whether merchants require those identity layers to be in place before participating, or discover their absence after adoption.

The Terms of Adoption

None of this argues against UCP adoption. The protocol solves real problems. Merchants need discoverability in AI interfaces. Standardised checkout reduces friction. Interoperability with payment rails simplifies implementation.

The argument is for adoption on terms that preserve merchant interests.

The parties who built UCP aren’t adversaries. Google is responding to genuine user demand for AI-mediated commerce. Payment networks are building infrastructure for an emerging market. The protocol itself is well-designed for its stated purpose.

Merchant interests simply aren’t identical to platform interests, and the terms of participation matter. Hotels that negotiated rate parity and direct booking capabilities with OTAs fared better than those who accepted default terms. Restaurants that maintained their own delivery and customer data alongside platform participation built more defensible businesses.

Merchants adopting UCP should apply the same strategic lens. Participate for reach. Demand data parity. Build verification requirements into integration contracts. Retain the ability to compete on experience rather than becoming interchangeable fulfilment infrastructure.

Building the Infrastructure

At AstraSync, we’ve been building the verification layer that UCP explicitly leaves to others. Our Commerce Shield product addresses the data parity gap: agent identity verification using the same KYC standards applied to financial counterparties, contextual metadata passthrough for fraud signals, and audit trails that document the instruction chain when disputes arise.

We’re not the only ones working on this problem. The infrastructure for data parity in agentic commerce is emerging across multiple providers. What matters for merchants isn’t which solution they choose. What matters is that they require a solution as a condition of UCP participation.

The window for negotiating these terms is now, while UCP adoption is early and merchants have leverage. Once agent-mediated commerce becomes normalised, the terms will calcify. The merchants who demanded protections will have them. Those who adopted on default terms will discover what they traded for discoverability.

The Paradox Resolved

Agentic commerce is coming regardless of what any individual merchant decides. The question isn’t whether to participate. It’s whether to participate as a strategic player or as infrastructure.

UCP adoption makes sense. Adoption without data parity protections trades customer relationships for visibility. The protocol is sound. The coalition is aligned. The merchant interest is simply to ensure the terms of adoption preserve what makes their business defensible.

Demand data parity. Require bank-grade agent verification. Build audit capabilities into integration contracts. Then adopt with confidence.

The merchants who get this right will gain the discoverability benefits of agentic commerce while retaining the customer intelligence that lets them compete. Those who don’t may find they’ve become the warehouses and logistics networks for platforms that own the customer relationship they used to have.

Further reading on AstraSync

• UCP: Universal Commerce Protocol
• Verify before transact: the trust pattern for agent commerce
• Merchants: start here

This essay first appeared on Medium on 13 January 2026.