India: UPI Prepares for the Arrival of AI-Driven Payments
India is gearing up for a new evolution of its instant payment infrastructure. A system could allow artificial intelligence agents to automatically carry out certain low-value transactions on UPI, without requiring human validation for each operation. The project may be presented soon in Mumbai.
The Indian digital payment system UPI could take a new step with the introduction of agent-based payments, allowing AI agents to autonomously perform certain operations.
According to several sources close to the matter, authorities and industry players are working on a framework that would allow these agents to conduct small transactions without requiring user confirmation for each one.
If the project materializes, India would be among the first countries to integrate AI-driven payments directly into a national payment infrastructure.
A New Generation of Digital Payments
The idea is based on a change in logic: instead of using an app to manually make each payment, a user could authorize an intelligent agent to act according to predefined rules.
The initial use cases would likely involve frequent and low-value purchases, such as grocery shopping. The agent could, for example, carry out certain recurring transactions within limits set by its user.
This evolution could gradually transform the way consumers interact with digital financial services.
UPI, an Already Massive Infrastructure
The project would rely on Unified Payments Interface (UPI), the instant payment system managed by the National Payments Corporation of India.
UPI already plays a central role in retail payments in India. According to data cited by the International Monetary Fund, it ranked in 2025 among the largest global networks for rapid retail payments by transaction volume.
The scale of the processed flows gives particular significance to the project. In August, UPI recorded 24.51 billion transactions, representing a total value of 29.820 trillion Indian rupees, or about 314 billion dollars.
Two major players dominate a large portion of this activity: Google Pay and PhonePe, which together account for about three-quarters of monthly volumes.
A Dedicated Protocol in Preparation
Efforts are also focused on a Unified Agent Protocol, which could be presented as early as next week during the Global Fintech Fest held in Mumbai.
The protocol aims to establish the rules allowing AI agents to interact with the payment ecosystem while regulating their ability to trigger transactions.
However, the definitive details remain confidential at this stage and may still evolve before a potential public presentation.
The Question of Payment Limits
One of the main questions concerns the protective mechanisms currently used by banks for automated payments.
According to the cited sources, some limits may be reexamined, particularly the provision for a cap of 10,000 rupees and a maximum period of 90 days for certain debit blocks.
Adapting these rules would be important to allow AI agents to conduct transactions flexibly while maintaining safeguards against errors, abuses, or unauthorized operations.
India Aims to Take a Stand in the Race for Agent-Based Payments
Several payment companies are already developing solutions that allow AI agents to intervene in transactions in the United States, Europe, Singapore, and Australia.
However, India's ambition would be to go further by directly integrating this technology into a national infrastructure used daily by hundreds of millions of people.
-
19:00
-
18:45
-
18:33
-
18:15
-
17:51
-
17:35
-
17:17
-
17:00
-
16:42
-
16:21
-
16:14
-
16:05
-
15:47
-
15:33
-
15:32
-
15:29
-
15:16
-
15:15
-
15:11
-
15:07
-
15:00
-
14:43
-
14:41
-
14:38
-
14:22
-
14:15
-
14:05
-
13:43
-
13:43
-
13:40
-
13:39
-
13:25
-
13:24
-
13:19
-
13:10
-
13:09
-
13:04
-
13:01
-
12:56
-
12:50
-
12:48
-
12:35
-
12:18
-
11:58
-
11:41
-
11:38
-
11:25
-
11:11
-
11:10
-
11:06
-
11:05
-
11:01
-
10:47
-
10:30
-
10:15
-
10:08
-
10:00
-
09:55
-
09:49
-
09:42
-
09:36
-
09:28
-
09:21
-
09:15
-
09:06
-
09:05
-
08:51
-
08:47
-
08:41
-
08:32
-
08:26
-
08:16
-
08:15
-
08:00
-
07:42
-
07:41
-
07:31
-
07:20
-
07:15
-
07:08
-
07:08
-
07:05