Ant International launches open-source AI payments protocol

Ant International launched the open source Agentic Mobile Protocol on April 28 in Kuala Lumpur to let mobile wallets and apps support AI agent payments, as most existing systems still rely on card networks.

The framework includes tools for delegating payment authority to AI agents, settling very small agent-to-agent transactions, and verifying agent identity.

Ant said it had open sourced the protocol and is working with more than 40 Alipay+ wallet partners that together cover 1.8 billion user accounts and 150 million merchants.

Juniper Research estimated digital wallet users reached 4.4 billion in 2025 and will top 6 billion by 2030.

The company also said it is piloting AI agent card transactions with Mastercard and Visa and is working with Google on related protocols.

🔗 Source: Ant International

🧠 Food for thought

Implications, context, and why it matters.

This protocol fits into a wider effort to create a shared language for AI commerce

Ant’s agentic commerce work sits within Google’s Universal Commerce Protocol, which aims to help AI agents, retailers, and commerce systems work together through the shopping process 1.
Shopify, Target, and Walmart back the effort, which suggests broader support across retail 1.
Ant International is also working with Google on the Agent Payments Protocol (AP2), which sets rules for AI agents making payments. Ant brings experience in alternative payment methods (APMs), including digital wallets and bank transfers, plus ties to wallet partners 2.
The Model Context Protocol (MCP) supports part of this ecosystem by linking AI systems with outside tools and services. Visa has described it as a standard layer for agentic commerce 3. Ant International’s Antom, the company’s merchant payment business focused on digitisation, said its agentic payment product uses MCP. The code is open-sourced on GitHub 4.

Agentic payments could add a new automated layer to the global economy

Inside businesses, these tools could streamline routine work. Ant International said its AI model Falcon forecasts trading volumes. It also said Falcon processed US$1.5 trillion in transactions in 2025, reached more than 90% accuracy, and cut foreign exchange (FX) costs by 60% 5.
Autonomous financial agents may also raise security demands. Ant International said its AI models identify high-risk transactions with 95% precision and catch deepfakes, which are AI-generated fake images, audio, or video, at a rate above 99.8% 5.
Smaller firms could gain access to tools that were harder to afford before. Antom’s AI app for small and medium-sized businesses, EPOS360, includes an AI copilot that helps with jobs such as setting up online stores and tracking cash flow 5.