Key Takeaways
- Highnote, which unifies issuing, acquiring, credit, ledger, and money movement on a single platform, has launched Agentic Commerce capabilities for AI-initiated payments, built on Visa Intelligent Commerce
- As agents begin initiating payments on behalf of consumers and businesses, a layer that handles execution, control, and authorization is taking shape on the issuer and merchant side of the rails
- E-commerce and payments teams should reframe agent readiness not just as a merchant-front experience, but as a backend question of credentials, tokens, and authorization rules
Highnote and Visa Unveil Infrastructure for AI-Initiated Payments
Courtesy of a partnership with Visa, modern issuing, acquiring, credit, ledger, and money movement platform Highnote has launched new Agentic Commerce capabilities this week.
finovate.comOn May 27, 2026, Highnote, which integrates issuing, acquiring, credit, ledger, and money movement into a single platform, announced Agentic Commerce capabilities in partnership with Visa. This is infrastructure that lets businesses securely run payments initiated by AI agents, so-called AI-initiated payments. According to the company's official announcement, the capabilities are built on Visa Intelligent Commerce.
What stands out is that this is not about the look and feel of chatbots or consumer-facing agents. Highnote owns the underlying mechanics, the credentials, tokens, and authorization, that fire when an agent initiates a transaction. Discussion of agentic commerce tends to fixate on the shopping experience, but this announcement addresses the issuing and acquiring rails that support its roots.
Inside Highnote's "Programmable Payments"
The core of these capabilities is giving AI agents the ability to pay while ensuring they can only act within rules set by the business. According to Highnote, software can initiate and execute transactions within predefined rules, spend controls, and approval structures. Rather than granting agents freedom, the design centers on clearly delimiting what is permitted.
In concrete terms, three things combine: provisioning programmable payment credentials, applying real-time authorization logic, and orchestrating execution across AI-initiated transactions. By pairing Visa's tokenization with Highnote's unified platform, the system enables dynamic, per-transaction authorization decisions without exposing raw card numbers. CEO John MacIlwaine framed it this way: "AI is quickly moving from insight to action. Businesses want AI to do more than recommend or analyze. They want it to initiate and execute real financial workflows." The challenge, he noted, is making those transactions "secure, controlled, and operational at scale."
The connection method matters too. Highnote connects to Visa Intelligent Commerce through Visa's Intelligent Commerce Connect, enabling agents to initiate and merchants to accept agentic transactions through a single integration. Bridging the issuer side and the merchant side with one connection is the implementation crux of this offering.
The First Use Cases Start with B2B Spend Management
The initial use cases Highnote highlighted were not consumer shopping but B2B operations. Invoice and accounts payable automation, vendor payments, operational spend management, and AI-assisted procurement, the everyday payment workflows of businesses, are the targets. The framing is about moving from manual purchasing processes to automated execution.
This choice makes sense. B2B payments are repetitive, rule-bound, and already wrapped in approval flows, making them easy to delegate to agents while delivering large economic upside from automation. Visa's Ivy Lee, VP and Head of Agentic for CMS, said that through Visa Intelligent Commerce, "we're enabling B2B workflows where agents can initiate and complete transactions at scale," emphasizing that businesses and developers can focus on differentiated experiences rather than payments.
The capabilities also look ahead to a broader sweep of AI-driven financial operations: intelligent procurement, dynamic payment routing, supplier optimization, recurring operational spend, and industry-specific autonomous payment experiences. The important thing is not to read these as a flat feature list but as expressions of one large shift, where payments that humans once approved one by one move to payments that software executes according to rules.
Why Issuer and Merchant Infrastructure Matters
Coverage of agentic commerce tends to skew toward the agent side, namely which AI a consumer shops with. But for a transaction to actually settle, the side that accepts, authorizes, and clears the payment must be prepared. Highnote's announcement is a concrete sign that this acceptance-side infrastructure is coming together.
Behind this is the standardization Visa is driving for agent payments. In October 2025, Visa introduced the Trusted Agent Protocol with more than 10 partners. It is an open framework designed on existing web infrastructure that helps merchants distinguish malicious bots from legitimate AI agents acting on behalf of consumers. The premise is that before agents can initiate payments, merchants first need to be able to identify what counts as a trusted agent.
On top of that, in April 2026 Visa announced Intelligent Commerce Connect. Positioned as a network-, protocol-, and token-vault-agnostic on-ramp, it lets merchants accept payments from multiple agent protocols, including the Trusted Agent Protocol, Machine Payments Protocol, Agentic Commerce Protocol, and Universal Commerce Protocol. Highnote was among the pilot partners, and this announcement sits squarely in that lineage.
In other words, Visa is assembling a layered structure: a standard to identify agents (Trusted Agent Protocol) and an on-ramp to unify diverse protocols (Intelligent Commerce Connect), on top of which issuing and acquiring platforms like Highnote load concrete payment capabilities. Payments in the agent era are emerging not as a single service but as a stacked architecture of standards, connection, and execution.
How Major Players Are Approaching Agent Payments
| Player | Role / Position | What it provides for agent payments |
|---|---|---|
| Highnote | Unified issuing, acquiring, credit, ledger, money movement | Programmable credential provisioning, real-time authorization, execution orchestration for AI-initiated transactions |
| Visa | Card network and payments infrastructure | Standardization and an on-ramp via Visa Intelligent Commerce, Trusted Agent Protocol, and Intelligent Commerce Connect |
| Merchants / E-commerce | The acceptance side | Accepting agent transactions, structuring product data, designing authorization policies |
What this table shows is that agent payments cannot be completed by any one company. Visa supplies the standards and on-ramp, Highnote handles the issuing and acquiring execution layer, and merchants accept the result. Because the roles are split, it is worth knowing which layer your own business depends on.
What E-commerce and Payments Teams Should Do Now
The top priority to check is how your issuers, acquirers, and payment partners plan to support agent-initiated transactions. Highnote's announcement signals that the issuing and acquiring infrastructure has started moving, and it is worth confirming early whether your payment stack is structured to accept and authorize future agent transactions.
Next, businesses with B2B payment workflows have room to begin exploring automation in repetitive spend areas like procurement and accounts payable. These are precisely the rule-bound domains with existing approval structures that Highnote chose as initial use cases. Because they are easy to delegate to agents and promise economic benefits, it is worth taking inventory of which of your payment processes suit "rule-based execution."
Beyond that, designing authorization policies for agent transactions is a new consideration to keep in mind. How you define spend limits, approval structures, and the scope of permitted agents is a design task that traditional payment operations never required. Together with structuring product data and tokenizing checkout, this is preparation that will create an edge as agent-driven purchasing takes off.
Conclusion
The Highnote-Visa partnership pulls the agentic commerce conversation back from the consumer shopping experience toward the issuing and acquiring infrastructure that supports it. Programmable credentials, real-time authorization, and tokenized transactions form the foundation for AI agents to initiate payments safely.
What to watch next is how far the Intelligent Commerce Connect pilot expands across partners, how large the B2B automation use cases grow in real transactions, and how Visa's standards like the Trusted Agent Protocol connect with other networks' frameworks. For e-commerce businesses, supporting agent payments is becoming a topic to factor in this quarter, not just as a front-end experience but as a backend question of credentials, tokens, and authorization.
