India Set to Introduce AI-Powered Agentic Payments on UPI, Sources Say

India is preparing a new framework that could allow artificial intelligence agents to make small digital payments on behalf of users without requiring approval for every transaction, according to three people familiar with the development. The move could place India's Unified Payments Interface (UPI) among the world's largest payment networks to support so-called agentic payments, where AI systems can carry out transactions based on instructions, spending limits and conditions previously set by customers.

Payment companies in the United States, Europe, Singapore and Australia have already been developing similar systems. However, introducing the technology through UPI would put India among the first countries to bring agentic payments to a nationwide digital payment infrastructure. UPI, operated by the National Payments Corporation of India (NPCI), has become the world's largest retail fast-payment system by transaction volume, according to a 2025 report from the International Monetary Fund. In August, UPI processed 24.51 billion transactions with a combined value of ₹29.82 trillion, or about $314.21 billion. Google Pay and Walmart-owned PhonePe accounted for nearly three-fourths of the monthly transaction volume.

The proposed Unified Agent Protocol is expected to be unveiled at the Global Fintech Fest in Mumbai next week, the sources said. They requested anonymity because the details of the plan have not yet been made public. NPCI had not immediately responded to requests for comment. In its early stages, the technology is likely to focus on frequent, low-value purchases. Grocery shopping could become one of the first practical applications, allowing an AI agent to complete routine purchases without asking the user to manually approve every payment. E-commerce companies could also become early adopters as they look to use AI to automate parts of the shopping process.

The potential applications could eventually go beyond routine purchases. Users could, for example, instruct an AI agent to buy a product whenever it becomes available at a particular discount or falls below a specified price. Similar rules could potentially be used for investments, allowing an agent to act when predetermined price conditions are met. NPCI is expected to build the new system around existing UPI features, including UPI Circle and Reserve Pay. UPI Circle enables a primary account holder to delegate payment authority to another user, while Reserve Pay allows customers to set aside funds for multiple future debits.

Banks currently place limits of up to ₹10,000 on such reserved funds for periods of as long as 90 days. These limits and the validity period could be reconsidered if the mechanism is expanded for AI-agent payments, according to the sources. Under the proposed framework, merchants would be able to integrate directly with NPCI's infrastructure, allowing customers to create rules governing when an AI agent can make payments and how much it can spend. Spending limits, identity verification and transaction records are expected to form part of the system's safeguards.

NPCI is also understood to be working on a liability framework to determine responsibility when problems occur, although details about how that system would operate have not yet been disclosed. The development comes as international payment giants Mastercard and Visa are also working on agentic payment technology in India. Their efforts form part of a wider global race to establish systems that allow AI agents to participate directly in digital commerce.

Mastercard completed its first authenticated agentic transaction in New Delhi in June. Earlier this year, fintech company Pine Labs launched its P3P protocol, which allows AI agents to complete UPI payments after receiving a single authorisation from the customer. The arrival of agentic payments could mark the next major stage in the evolution of UPI. The technology could make everyday transactions more automated, but its wider adoption will depend on strong safeguards that prevent unauthorised spending and give users clear control over what AI agents can and cannot do with their money.

Disclaimer: This image is taken from Reuters.