Three of South Africa's largest banks have quietly joined the most ambitious attempt yet to solve a 30-year-old problem: giving AI agents the ability to pay for things without human intervention. FNB, Absa, and Nedbank are now members of Open Standard, a consortium of over 140 companies including Visa, Mastercard, Stripe, Google, and Amazon Web Services, working to launch Open USD (OUSD), a dollar-backed stablecoin designed specifically for machine-to-machine transactions.

The timing reveals something important about where AI commerce is heading. AI agents can already research suppliers, analyze financial data, and plan complex logistics chains without human supervision. But the moment money needs to change hands, the internet's infrastructure breaks down. Agents hit a wall of card numbers, one-time passwords, subscription logins, and checkout pages, all designed assuming a human is present. Open Standard and its partner protocol, x402, are designed to tear down that wall.

How Does Machine-to-Machine Payment Actually Work?

The solution involves two complementary pieces of infrastructure working together. The first is x402, an open payment protocol launched by Coinbase in May 2025 and joined by Cloudflare that September. It repurposes HTTP status code 402, "Payment Required," a placeholder reserved in the web's specifications since the 1990s but never actually used.

Here's how the exchange works in practice:

• Agent Request: An AI agent requests access to a paid API, dataset, or article from a server.
• Price Quote: The server responds with a 402 status code containing the price for access.
• Payment Decision: The agent evaluates the cost against its budget and decides whether to proceed.
• Stablecoin Transaction: If approved, the agent signs a micro-transaction, typically in Circle's USDC stablecoin, and attaches proof of payment.
• Instant Access: The entire exchange completes in a single HTTP round trip, with no subscription, account creation, or human involvement required.

The usage numbers show how different this economy looks from the one built for humans. The x402 protocol processed approximately 75 million payments in the past 30 days, moving US$24 million, with an average payment of just $0.32. No traditional card network can process a charge that small profitably. x402 exists for the payments the card rails cannot carry.

Why Do AI Agents Need Their Own Currency?

A communication protocol alone is only half the puzzle. For agents to transact at enterprise scale, they need a neutral, frictionless currency to settle in. Until now, that has largely meant USDC, issued and controlled by Circle, a single company that charges tiered redemption fees on large volumes.

Open USD fills that gap. Announced on June 30, the consortium designed OUSD as shared infrastructure rather than a proprietary product. Minting and redemption are free at any volume, reserve income flows back to the partners that drive adoption, and governance sits with a board drawn from the membership rather than a single issuer. The coin is slated to go live later this year.

"It's a stablecoin built for the internet economy, designed by the businesses growing it," said Zach Abrams, Open Standard's founding CEO, whose previous company, Bridge, was bought by Stripe for $1.1 billion.

Zach Abrams, Founding CEO at Open Standard

Crucially, Open Standard explicitly lists "agentic commerce" among its core target use cases, alongside payments, remittances, and trading. The goal is to enable agents to "instantly make programmatic payments" without friction.

Why South Africa's Banks Are Positioning Themselves Now

The South African connection is not incidental. Visa has already enrolled South African banks in its Agentic Ready programme, and local institutions are investing heavily in agentic AI across their operations. Membership of Open Standard gives South African developers and fintechs a direct line to build agents capable of autonomous, cross-border micro-transactions in a stablecoin plugged into local banking infrastructure.

The ground is particularly fertile in Africa. According to DHL's E-Commerce Trends Report 2026, 56 percent of online shoppers in sub-Saharan Africa say they are likely to let AI make shopping decisions or even purchases on their behalf within the next five years, nearly double the global average of 29 percent. Africa is already among the world's fastest-growing markets for stablecoin adoption.

Settlement would happen in seconds on-chain rather than days through correspondent banking and Swift, though reserve bank exchange-control rules will still govern rand conversions at either end.

How Does This Compare to India's Approach?

Meanwhile, India is taking a different path tailored to its own payment infrastructure. The National Payments Corporation of India (NPCI) confirmed in July 2026 that it is building a Unified Agent Protocol (UAP), a framework that would let AI agents be registered, verified, and authorized to make payments over the Unified Payments Interface (UPI).

The scale of UPI is enormous. The system cleared over 22 billion transactions worth more than 28 lakh crore rupees (approximately US$296.92 billion) in June 2026 alone, up 23 percent in volume and 20 percent in value year on year, averaging 757 million payments a day. For fintech teams in India, the work starts now, before the protocol ships.

UAP would sit on top of the existing UPI rails rather than replace them, functioning as a trust and verification layer. An AI agent would be registered and given verifiable credentials, then checked at payment time to confirm it is a legitimate agent, that it is authorized to act for a specific user, and that the request falls inside the limits that user set.

The stated use cases are ordinary recurring life admin: paying a utility bill before its due date, renewing a subscription, buying a transport ticket, booking travel, or ordering groceries.

What Technical Challenges Must Every Agentic Payment System Solve?

Whether using x402 and OUSD or India's UAP, every agentic payment design must solve three core problems. The first is authorization. The system has to establish not just who the user is, but whether their agent is legitimate and is allowed to act for them. That means an agent needs its own identity and credential, distinct from the human's payment ID, so a payment can be tied to a specific agent instance rather than a shared secret.

The second is the scope of authority. Once an agent can pay, the user has to be able to say how much, how often, and for what. A protocol that only answers "is this agent real" without answering "what is it allowed to spend" is unfinished. This is where per-agent limits and category rules live.

The third is accountability. If an agent overspends, buys the wrong thing, or is compromised, someone has to be answerable, and the transaction has to be traceable and ideally reversible. NPCI's plan to hold logs of agentic transactions is a start, but liability between the user, the app, the agent provider, and the bank is still unsettled.

What Happened to Facebook's Libra, and Why Is This Different?

If a corporate mega-consortium launching a global currency sounds familiar, it should. In 2019, Facebook unveiled Libra with a similar cast, including Visa, Mastercard, and Stripe as founding members. The project unraveled under regulatory pressure, with the payments giants quitting within months. Renamed Diem in a bid for respectability, it sold off its assets and wound down in early 2022.

Two things have changed. First, there is no Meta at the center this time. Open Standard is structured so that no single company controls it, which was Libra's original sin in regulators' eyes. Second, the US now has a stablecoin law, the Genius Act, signed in July 2025, giving fully reserved, regulated dollar tokens a legal foundation Libra never had. The same firms that fled Libra are now building its successor inside the regulatory perimeter, not outside it.

"Bank accounts and traditional payment rails were not designed for machine intelligence, to say nothing of the regulatory burdens that create friction in traditional systems," said Simon Dingle, who helped launch ZARP, South Africa's first rand-pegged stablecoin.

Simon Dingle, Co-founder of ZARP

Dingle added that a machine cannot open a bank account, but it can set up a crypto wallet in seconds and begin transacting with stablecoins via protocols like x402. He drew a line, however, between crypto-native stablecoins like ZARP or USDT, which operate on open, public networks like Ethereum and Solana, and other instruments that use the same name and technology for something entirely different in substance, reintroducing friction from traditional finance into the digital payment space.

The genuine article lives in the open. Open Standard has said OUSD will initially run on public blockchains, ensuring that the infrastructure supporting agentic commerce remains transparent and accessible rather than locked behind proprietary systems.

" }