Payments technology has always competed on features. Real-time settlement, tokenization, fraud scoring, reconciliation automation: each was once a differentiator, and each became, within a product cycle or two, a commodity baseline. Artificial intelligence is compressing that timeline. What took five years to commoditize is now taking two, and the payments organizations that understood this dynamic before deploying AI will be structurally better positioned than those that treated AI as one more feature to ship.
From Aspirin to Vitamin: How AI Is Reclassifying Payments Capabilities
Judd Howard, Senior Product Manager of AI at Spreedly, offered a framework in a PYMNTS analysis published June 22 that explains the shift with unusual precision. Infrastructure functions such as tokenization, processing, and reconciliation have moved into the “aspirin” category: essential operational requirements that no longer differentiate. Meanwhile, workflow orchestration, advanced analytics, and AI-driven data services remain “vitamins” that optimize strategy without being necessary for basic operation.
The practical implication: payments organizations cannot use AI to win on the infrastructure layer anymore. That layer is getting commoditized faster than anyone can ship. The competitive surface has moved to trust, compliance network relationships, and operational resilience: the things AI can accelerate but cannot itself replicate from scratch. As Howard put it, “AI can replicate software faster than it can replicate trust.”
Mastercard’s Agent Pay for Machines Shows Where the Rails Are Going
Mastercard’s introduction of Agent Pay for Machines makes concrete what the theoretical shift looks like in practice. The product extends Mastercard’s existing Agent Pay framework specifically to machine-to-machine transactions: scenarios where software executes payments on behalf of other software, without a human reviewing each transaction in real time.
The use case Mastercard describes is precise. A small business owner delegates a task to an AI agent: build a website, launch a marketing campaign, manage payments across multiple funding sources. The agent needs to spend autonomously, across cards, virtual cards, bank balances, lines of credit, and potentially stablecoins, while maintaining verifiable records of what was authorized, by whom, and for what purpose.
Mastercard’s answer to the authorization problem is its Verifiable Intent Framework, which links identity, instructions, and outcomes in a way the company has shared with FIDO Alliance for industry-wide adoption. Sherri Haymond, Mastercard EVP, made the standardization point explicit: “Standardization is key. There are some things that just aren’t competitive. They’re foundational, enabling pieces.”
That framing echoes Howard’s aspirin/vitamin distinction. Mastercard is treating the identity and authorization infrastructure for agentic commerce as a commodity layer to be standardized, not a proprietary moat to be defended. The bet is that the value flows to whoever owns the trusted network layer underneath the commodity infrastructure.
What This Means for the Financial Services Leader
The signal from the earlier wave of agentic payment infrastructure announcements was that the pipes were being built. The June 22 set of developments suggests the pipes are ready and the question has shifted to who gets to control the flow through them.
For payments executives, three questions follow from this shift.
Where is your differentiation, and is it still a vitamin or already an aspirin? Organizations that built differentiation on capabilities AI can now replicate in a product cycle need a new answer. The honest audit is to ask whether the core offering is genuinely hard to replicate or merely familiar.
Who owns the trust layer in your agentic deployment? Mastercard’s approach of standardizing the Verifiable Intent Framework and sharing it with FIDO is a play for the trusted network position. Banks, processors, and payment platforms evaluating agentic commerce need to decide whether to adopt those standards or build competing ones. Most will not have the scale to win a competing standards battle.
How ready is your compliance and infrastructure for the agentic transaction volume? Howard’s observation that “payments infrastructure has been set up with a lot of the fundamentals around compliance and security” is an argument that organizations with strong pre-AI foundations may actually be better positioned for the AI transition than those that started building clean. If that is true, the incumbents have a window, but it will not stay open indefinitely.
The Real-Time Payment Parallel
The adoption arc of real-time payments, which crossed the commercial inflection point as B2B adoption intent reached 53 percent, offers a useful reference. The capability was technically available for years before it achieved the data quality, network density, and trust infrastructure required for mainstream commercial adoption. AI in payments is following a similar arc, but faster.
The organizations that built the data and trust foundations before the adoption curve steepened in real-time payments were able to move fast when the market arrived. The same dynamic is likely to play out in agentic commerce. The question for financial leaders is not whether agentic payments will be mainstream. It is whether the infrastructure being built now will be theirs or someone else’s.
Source: PYMNTS
