Ant International is trying to position itself at the infrastructure layer of agentic commerce by creating a standard that allows AI agents to initiate payments through existing digital wallets.
The company open-sourced its Agentic Mobile Protocol, or AMP, on GitHub, effectively inviting developers and payment providers to adopt a common framework for AI-driven purchases.
AMP is built on Ant International’s Alipay+ cross-border network. The initial phase connects to 10 major Asian digital wallets serving about 1.5 billion consumer accounts.
That scale gives Ant an important starting advantage. Rather than asking consumers to create a new payment method for AI agents, the protocol is designed to let agents operate within wallet systems that users already trust.
The participating ecosystem includes platforms such as Alipay in mainland China, AlipayHK in Hong Kong, Starryblu in Singapore, and KakaoPay and Toss in South Korea.
The difficult part is identity. A human consumer can authenticate through a phone, biometrics or account credentials. An autonomous software agent needs a different system for proving who authorized it, what it is allowed to buy and how merchants should verify that authority.
Ant is addressing that problem separately with Visa and Mastercard through a “know-your-agent” framework. The goal is to create standardized identity verification for AI agents across card networks, wallets and online marketplaces.
That work could become as important as the payment protocol itself. Agentic commerce will not scale if merchants and financial institutions cannot distinguish authorized software from fraud.
The opportunity is large because technology companies are moving toward agents that can complete entire commercial tasks rather than simply recommend products. An agent could compare options, choose a product and execute payment without requiring the user to manually complete every step.
For Ant, the strategic value is standard-setting. If AMP becomes widely adopted, the company can influence how agentic payments interact with mobile wallets even outside its own direct consumer products.
Open-sourcing the protocol is part of that strategy. A closed standard can be easier to monetize directly but harder to establish globally. An open framework can spread faster and make Ant’s existing Alipay+ network more important to developers.
The main risks are security, regulation and consumer trust. Users will need clear controls over spending limits, merchant permissions and dispute resolution. Regulators will also need to decide who is responsible when an autonomous agent makes a mistake.
The race is therefore not simply to make AI capable of shopping. It is to build the identity and payment rails that make automated spending safe enough for financial institutions to accept. Ant is trying to make sure those rails run through infrastructure it helped define.
Another unresolved issue is liability. If an agent buys the wrong item, exceeds a user’s intent or is compromised, wallet providers, merchants and card networks will need clear rules for authorization, refunds and fraud responsibility.
Merchant adoption will determine whether AMP becomes a real standard. Wallet connectivity on the consumer side is useful, but agents also need merchants to recognize permissions, process refunds and handle disputes. Visa and Mastercard participation can help because their networks already connect millions of merchants and have mature fraud controls. If the KYA framework and AMP can work together across wallets and card rails, Ant could reduce the fragmentation that would otherwise slow agentic commerce. The value of the protocol will be measured by transaction volume, not GitHub adoption alone.