In our previous coverage of trust as a competitive battleground (Post 128003) and the commoditization of payment rails (Post 128411), we established that the infrastructure layer is rapidly becoming a utility. However, the industry’s current obsession with the intelligence of agentic commerce—the LLMs and the conversational interfaces—is largely a distraction. The actual bottleneck preventing the scaling of autonomous payments is not the ability of an agent to reason, but the inability of the current governance framework to account for its actions.
Thredd CEO Jim McCarthy has recently articulated a framework that shifts the focus from the interface to the ledger. McCarthy argues that the defining challenge for the industry is governance, specifically the triad of ‘permission, control and proof.’ Without these, an agent is merely a liability generator. The industry is currently attempting to force-fit autonomous agents into a manual, human-centric authorization flow, which is an exercise in futility.
The McCarthy framework proposes that the solution lies in ‘verifiable intent’ via cryptographic binding to credentials. In the current environment, ‘permission’ is a binary, static state. By moving to cryptographic binding, the industry can transition from simple permitting to active underwriting. When an agent’s intent is cryptographically bound to a specific set of credentials, the issuer is no longer guessing at the legitimacy of a transaction; they are verifying a signed instruction set. This is the difference between a blind authorization and a calculated risk decision.
This shift is necessary because the operational divide between current payment systems and agentic requirements is widening. We are currently operating under the assumption that agents can function within the existing rules of engagement. This is incorrect. The operational reality is that agents require a continuous, verifiable audit trail that current legacy systems are not architected to provide. The industry is trying to build a skyscraper on a foundation designed for a shed.
The most glaring evidence of this incompatibility is the current chargeback regime. Legacy chargeback rules were written for human consumers who can be held accountable for their actions. They are fundamentally incompatible with standing instructions issued by third-party agents. If an agent executes a transaction based on a misinterpreted prompt, the current dispute resolution process—which relies on human intent and merchant-consumer interaction—breaks down entirely. You cannot hold an algorithm liable in a court of law, nor can you easily reverse a transaction when the ‘intent’ was technically valid but contextually flawed.
Issuers who successfully implement a system to prove intent will be the ones who effectively underwrite agentic commerce. By moving the verification process to the point of origin, issuers can mitigate the risk that currently keeps them on the sidelines. This is not about making agents smarter; it is about making the liability ledger more robust. The real action is not in the chat interface where the user interacts with the agent, but in the backend where the intent is bound to the payment credential.
As noted in the PYMNTS eBook ‘Building the Agent-Ready Payments Enterprise’ (July 29, 2026), the transition to an agent-ready enterprise requires a fundamental re-engineering of how we define authorization. If the industry continues to prioritize the user experience of the agent over the governance of the transaction, we will see a proliferation of failed payments and unresolvable disputes. The market is currently waiting for a standard that allows for verifiable, cryptographically bound intent.
Ultimately, the hype surrounding agentic commerce will subside, leaving behind the boring, necessary work of governance. Those who focus on the mechanics of the liability ledger will find themselves in a position to capture the value that the current, fragmented system is leaking. The future of payments is not about better chatbots; it is about better proofs.
Sources:
1. PYMNTS eBook ‘Building the Agent-Ready Payments Enterprise’ (Jul 29, 2026).
2. Thredd CEO Jim McCarthy, ‘permission, control and proof’ framework.
3. Forkast Post 128003 (Paymentology analysis).
4. Forkast Post 128411 (Settlement Architecture).