India May Allow AI Agents to Make Small UPI Payments Without User Approval

India may soon allow AI agents to make small UPI payments without individual approval, under a proposed framework designed to enable secure, rule-based agentic transactions.

India May Introduce AI-Powered UPI Payments

India could be moving towards a new phase of digital payments in which AI agents may be authorised to complete small UPI transactions without requiring users to approve every payment individually.

According to a Reuters report citing people familiar with the matter, the proposed system is known as the Unified Agent Protocol. The framework could potentially be introduced at the Global Fintech Fest in Mumbai.

However, the proposal has not yet been officially confirmed by the National Payments Corporation of India (NPCI), which operates the UPI payment network.

What Is the Unified Agent Protocol?

The proposed Unified Agent Protocol could allow users to provide AI agents with limited, pre-authorised payment permissions.

Instead of asking a user to approve every individual transaction, an AI agent could make payments according to instructions and spending limits established in advance.

For example, a user could potentially instruct an AI agent to purchase groceries up to a specific amount. The agent would then be able to complete qualifying transactions without seeking separate approval each time.

The exact functionality, transaction limits and security requirements will depend on the final framework if NPCI officially introduces it.

Small and Frequent Payments Could Come First

According to the reported proposal, the initial focus could be on low-value and routine transactions.

Potential early applications could include:

• Grocery purchases

• Utility bill payments

• Routine subscriptions

• Cab bookings

• Small online purchases

• E-commerce transactions

AI agents could potentially search for products, compare available offers and discounts, select suitable options based on user instructions and complete eligible payments.

More sophisticated applications, including certain investment-related transactions based on predefined price conditions, could potentially emerge later, subject to regulatory approval and system design.

How UPI Infrastructure Could Support AI Agents

The reported framework could build on existing UPI capabilities instead of creating an entirely separate payment system.

One possible component is UPI Circle, which allows users to delegate payment authority to another person. Another is Reserve Pay, which can enable funds to be set aside for specified payments.

NPCI could also reportedly provide infrastructure that allows merchants to connect their systems with AI-powered payment agents.

This could help create an ecosystem in which AI agents, consumers, merchants and payment infrastructure work together under defined rules.

Spending Limits and Security Measures

Allowing AI systems to make payments independently would require strong safeguards. The reported framework could therefore include multiple controls to reduce the risk of unauthorised transactions.

Potential safeguards include:

• Predefined spending limits

• Identity verification

• Transaction monitoring

• Audit trails

• User-defined payment rules

• Liability mechanisms

These controls would allow users to determine how much an AI agent can spend, where it can make payments and under what conditions.

The proposed approach would therefore be different from simply giving an AI unrestricted access to a bank account. Instead, the agent would operate within a controlled payment authority established by the user and payment system.

Expert View on Agentic Payments

Rohit Mahajan, Founder and CEO of plutos ONE, described agentic payments as a potential next step for UPI.

He suggested that AI agents could eventually handle routine financial activities such as paying bills, managing subscriptions, ordering groceries, booking transportation and completing small purchases within consumer-defined limits.

Such capabilities could make digital payments more automated, allowing users to delegate repetitive transactions to AI systems while retaining control through predefined rules.

Global Payment Networks Are Exploring Agentic Commerce

India is not alone in exploring AI-driven payments. Major global payment networks and fintech companies are also developing infrastructure for agentic commerce.

Mastercard has developed Agent Pay, which is designed to support transactions conducted by AI agents while incorporating controls for security, trust and transaction visibility.

Visa is also working on infrastructure aimed at enabling AI agents to participate in commerce. Such systems could allow an agent to search for products, make decisions based on user preferences and initiate purchases while maintaining authentication and security controls.

Indian companies are also exploring this emerging segment, with businesses including Pine Labs working on infrastructure related to AI-powered commerce and payments.

Why AI-Powered UPI Payments Could Be Significant

UPI's enormous scale makes the proposed development particularly important. The payment network processed 24.51 billion transactions worth ₹29.82 lakh crore in August, highlighting the size of the ecosystem in which agentic payments could potentially operate.

If successfully implemented, AI-powered transactions could make routine digital payments faster and more automated. They could also create new opportunities for e-commerce platforms, merchants, banks and fintech companies.

However, widespread adoption would depend heavily on security, consumer protection, accountability and regulatory clarity.

NPCI Has Yet to Confirm the Proposal

The Unified Agent Protocol remains a reported proposal rather than a formally confirmed NPCI framework.

Important details, including transaction limits, eligibility requirements, authentication procedures, merchant integration, liability rules and rollout plans, will depend on any official announcement.

Until then, users should treat reports about AI agents making autonomous UPI payments as a potential upcoming development rather than an already available UPI feature.

Conclusion

India's reported Unified Agent Protocol could mark an important evolution in digital payments by allowing AI agents to complete small, rule-based UPI transactions without individual approval. If introduced with strong safeguards, the framework could bring greater automation to everyday payments while keeping users in control.