Agentic commerce promises frictionless, AI-driven purchasing experiences. But beneath that promise lie questions about responsibility, one that merchants are only beginning to understand.
As AI agents begin making purchases autonomously, proving trust becomes far more complex. Authorization alone is no longer enough. Merchants must be able to verify agent identity, preserve user intent, and maintain a defensible chain of context across multiple protocols, providers, and payment networks.
To address this emerging challenge, IXOPAY and ZIP have introduced a joint initiative to define a Unified Trust Layer—an open framework that orchestrates identity verification, intent capture, and behavioral trust signals into a portable trust artifact that travels with the transaction. By combining IXOPAY’s merchant-owned tokenization and orchestration with ZIP’s real-time behavioral intelligence, the initiative aims to give merchants the infrastructure needed to defend and scale agent-driven commerce.
We sat down with IXOPAY CTO Jill Willard to unpack what’s really changing—and why merchants must evolve their payments infrastructure now.
Q: Agentic commerce is often described as frictionless and transformative. What’s the hidden reality for merchants?
Jill Willard:
The promise is compelling: AI agents autonomously purchasing on behalf of consumers, delivering hyper-personalized and seamless experiences.
But what no one is talking about enough is the liability shift.
When an AI agent makes a bad purchase—whether it’s fraud, misuse, misunderstood intent, or simply an error—the merchant still carries the risk. This could look like the velocity of transactions (think rogue agent), or dispute and chargeback risk. Current frameworks weren’t designed for non-human actors. They don’t explicitly account for agent-initiated transactions.
So while customers gain convenience, merchants assume greater responsibility—without the infrastructure to prove identity, intent, and consent in a defensible way.
That’s the hidden reality.
Q: Why isn’t existing payments infrastructure enough?
Jill:
Traditional infrastructure was built around a four-party risk model—consumer, merchant, issuer, and acquirer, assuming a human initiated the transaction.
Agentic commerce breaks that assumption.
Today’s systems focus heavily on routing, optimization, and reconciliation. Even payment orchestration layers were not designed to preserve machine-verifiable identity, scoped consent, behavioral context, and policy constraints as portable trust artifacts.
When trust signals are scattered across agentic protocols—whether from Amazon, Google, Visa, or others—merchants are left stitching together fragmented data. In addition, while payment platforms and merchants are comfortable verifying identity, there is no mechanism for validating not just identity but also behavior. Is an agent behaving in a way that is expected?
You need replayability and trust verification.
Q: What do you mean when you say replayability and trust verification?
Jill:
In an autonomous environment, the audit trail becomes the merchant’s primary legal and financial defense against chargebacks and fraud. This means being able to replay not just the transaction but the elements leading up to the transaction.
Merchants must be able to document:
Who or what initiated the transaction
The scope of consent
The intent behind the purchase
The chain of custody across providers
Without that, defending against disputes is nearly impossible.
And here’s the bigger issue: Even if you have “intent,” replayability, and a valid agent identity, you’re still dealing with a probabilistic system.
We’ve all been there. You ask an agent for something simple, and it confidently gives you the wrong thing. Not malicious—just wrong.
And the worst part? You can’t see why it made that decision. So you can’t actually trust the process. You can only evaluate the output.
That’s where a Trust Score comes in.
It doesn’t rely on what the agent thinks it’s doing. It measures what happened against external rules that the agent doesn’t even know exist.
Because the hallucination risk is real.
These models are optimized to be helpful and convincing — not necessarily correct. They will confidently tell you something false and make it sound right.
Automatic trust in that world equals automatically accepting a lie as truth.
A Trust Score flips that model.
It gives the merchant insight and confidence in the model's output before it reaches the payment layer—grounding decisions in facts, not confidence.
The infrastructure must evolve from simply moving money to preserving proof and validating trust.
Q: How does tokenization evolve in this new model?
Jill:
Tokenization is no longer just about protecting a PAN for PCI compliance. In agentic commerce, it expands to capture the full context of a transaction—including the payment credential, agent identity, scoped consent, and user intent.
Instead of a single token, you now have a composite, verifiable token structure in which these elements are bound together and replayable. This enables end-to-end transaction reconstruction for auditability and dispute defense.
Frameworks like Visa TAP and Mastercard Verifiable Intent start to push intent into the authorization flow via Network Token cryptograms, but the real shift is persisting and binding that intent to the payment itself.
In a Trust Framework model, all of this is unified into a single token construct, including the Trust Framework's unique behavioral trust score, so that downstream systems can use the score for real-time risk and authorization decisions.
Tokenization evolves from protecting data to proving trust.
Q: This is where the Unified Trust Layer comes in?
Jill:
Yes, exactly.
Together with ZIP, we’re developing an open framework that embeds transparency, auditability, and interoperability into agent-driven transactions.
The Unified Trust Layer focuses on three core capabilities:
Agent Identity Validation: Normalizing identity verification across fragmented agentic protocols.
Intent Capture and Preservation: Binding explicit human consent and transaction context into an immutable, replayable record.
Contextual Trust Signaling: Aggregating behavioral signals to support informed authorization decisions and strengthen dispute defensibility.
IXOPAY’s merchant-owned, provider-agnostic tokenization and orchestration, combined with ZIP’s real-time behavioral intelligence, provide a unique foundation for the community to help build a Trust Framework that assists merchants in navigating the evolving risk landscape of agentic commerce.
This moves orchestration from routing payments to orchestrating trust.
Q: You’ve described this as the evolution from payment orchestration to trust orchestration. What does that look like for merchants?
Jill:
Payment orchestration optimizes transactions. Trust orchestration protects them.
In practical terms, that means:
Immutable, detailed audit trails for every agent-driven transaction
Aggregating trust signals across identity, consent, and behavior
Wrapping those signals into interoperable, tokenized artifacts
Providing a single integration layer across every agentic protocol
Merchants shouldn’t have to build separate pipes for Google’s protocol, Amazon’s, or anyone else’s. They need one infrastructure layer that preserves trust across all of them.
That’s the required evolution.
Q: What tangible value does this bring to merchants?
Jill:
Three immediate benefits:
Reduced chargeback exposure through comprehensive, defensible audit trails.
Fewer false declines and fraud losses because trust signals are aggregated and measurable.
The confidence to embrace new commercial models without exposing the business to uncontrolled risk.
Q: Why make this an open initiative?
Jill:
Trust can’t be proprietary in a networked ecosystem.
If agentic commerce is going to scale, identity and intent must function consistently across platforms, networks, and providers. That requires collaboration.
We’re inviting merchants, networks, and ecosystem participants to help define how trust should operate in AI-initiated commerce.
Merchants who prepare their trust infrastructure today won’t just survive this shift, they’ll lead it.
Q: Mastercard recently announced its Verifiable Intent protocol for AI-initiated transactions. How does the Unified Trust Layer compare?
Jill:
It’s an important development, and it reinforces something we’ve been saying for a while: as commerce becomes more autonomous, trust has to become explicit and verifiable. The industry is clearly recognizing that agentic transactions require new infrastructure.
Mastercard’s Verifiable Intent focuses on creating cryptographic proof that a consumer authorized an AI agent to act on their behalf. That’s a critical step, because proving user authorization and preserving intent will be foundational to agent-driven commerce.
Where the Unified Trust Layer expands on that concept is in how those trust signals are operationalized for merchants across the broader ecosystem.
At IXOPAY, we see intent verification as one of several signals that must be orchestrated together. Through the Unified Trust Layer, intent tokens become part of a broader trust artifact that also includes identity verification, behavioral signals, and contextual transaction data. Those elements can be aggregated and analyzed to provide merchants with deeper insights into agent behavior and transaction legitimacy.
Over time, we also expect to incorporate network-level trust signals into that model. For example, data associated with protocols like Verifiable Intent can be ingested into the behavioral intelligence layer alongside our own signals and merchant data, enriching the trust layer and providing merchants with a more complete view of risk.
In that sense, initiatives like Mastercard’s are complementary. They help establish foundational trust primitives, while the Unified Trust Layer focuses on an additional layer of behavioral validation and orchestrates those signals across providers, protocols, and payment networks so merchants can operationalize trust in real-world commerce.
If anything, these announcements validate the need for a broader trust infrastructure as agentic commerce scales.
The Future of Commerce Requires Trust Infrastructure
Agentic commerce is here. Consumers are already using LLMs to do shopping research. It won’t be long before agents are purchasing autonomously. The liability shift is real. And merchants cannot afford to rely on infrastructure built for a human-only world.
The role of payment orchestrators has expanded—they must now orchestrate trust.
With the Unified Trust Layer initiative, IXOPAY and ZIP are providing the foundations to build what trust looks like across this new landscape. Interested in joining the movement? Find out more by registering for our informational webinar.