India plans agentic payments on UPI, allowing AI agents to make transactions with preset limits: Report

India is preparing a framework that could allow artificial intelligence agents to make small digital payments on behalf of users without requiring approval for every transaction, Reuters reported, citing people familiar with the matter.

The proposed system would be built on the Unified Payments Interface (UPI), potentially making India one of the first countries to deploy agentic payments through a national digital payments infrastructure, the report said.

The framework, called the Unified Agent Protocol, is likely to be unveiled next week at the Global Fintech Fest in Mumbai, the report said, adding that under the proposed system, users would be able to give AI agents rule-based instructions on when and how much they can spend. The infrastructure is expected to include spending limits, audit trails and identity checks.

The move comes as payment companies globally work on systems that allow AI agents to carry out transactions on behalf of consumers. Companies have already introduced or are developing similar frameworks in the US, Europe, Singapore and Australia.

How AI agents could pay through UPI

The initial use cases in India are expected to involve frequent, low-value purchases such as groceries, Reuters reported.

E-commerce platforms could be among the early adopters. Instead of manually approving every transaction, a consumer could authorise an AI agent to make purchases within predefined rules.

More from Business

The system could eventually support more sophisticated tasks. An AI agent, for instance, could identify a product being offered at a discount and place an order automatically. It could also potentially make investments based on instructions such as specified price thresholds, according to the report.

The proposed framework is expected to build on two existing UPI mechanisms: UPI Circle and Reserve Pay.

UPI Circle allows a primary account holder to delegate payment authority to another user. Under the proposed agentic model, such delegated authority could potentially be extended to an AI agent.

Reserve Pay, meanwhile, allows customers to block funds for multiple debits. Banks currently cap such blocks at Rs 10,000 for up to 90 days, although those limits and the validity period could be revisited for agentic payments, the report said.

UPI gives India a large base for agentic commerce

The scale of UPI gives India a potentially significant advantage as AI-driven payments develop.

UPI is the world’s largest retail fast-payment system by transaction volume, according to a 2025 International Monetary Fund report. It processed 24.51 billion transactions worth Rs 29.82 trillion ($314.21 billion) in August.

Google Pay and Walmart-owned PhonePe together accounted for around three-fourths of UPI transaction volumes during the month.

That existing scale means an agentic payment layer could potentially be introduced on top of an infrastructure already used by hundreds of millions of consumers and merchants, rather than requiring an entirely new payments network.

NPCI is expected to provide infrastructure that merchants can integrate directly. This would allow customers to establish rules governing an AI agent’s payment activity, including spending limits and other controls.

According to the report, NPCI plans to establish a liability framework for the system, although details were not immediately available.

Mastercard, Visa also developing agentic payments

India’s move comes as global payment networks and fintech companies race to establish standards for AI-driven commerce.

Mastercard and Visa are separately developing agentic payment capabilities for the Indian market, Reuters reported. Mastercard completed its first authenticated agentic transaction in New Delhi in June, according to the report.

Fintech company Pine Labs has also entered the space. Earlier this year, it launched P3P, an agentic payment protocol that allows AI agents to complete UPI payments following a single upfront authorisation.