Artificial intelligence is moving beyond answering questions, writing code, and creating images. A new category of AI called AI agents has arrived. It is a kind of system which is designed to do tasks autonomously. Unlike regular AI chatbots that simply respond to prompts, AI agents can complete entire tasks. You could ask one to find the cheapest flight for you, make the booking, and even add the trip to your calendar. But in India, AI agents have so far faced one major limitation. They can browse, recommend, compare, and decide. But when it comes to making a payment, every AI agent hits the same wall — human authentication.
Today, UPI payments require a user to enter an MPIN, a step designed for humans, not machines. Pine Labs now says it has found a way around that problem.
The missing piece: Letting AI complete payments
The company says it has built a system called Pine Labs Payment Protocol (P3P) that allows AI agents to complete UPI payments without requiring human authentication for every transaction. The system is built on top of UPI’s existing payment infrastructure — UPI One Time Mandates (OTM) and Reserve Pay.
UPI One-Time Mandate was originally introduced to allow users to approve a future payment in advance. It temporarily blocks a fixed amount in a bank account for a single future transaction. This system is commonly used for services such as IPO applications and bookings.
Reserve Pay, powered by NPCI’s Single Block Multiple Debit framework, was designed to enable recurring or automatic payments. Users approve a spending limit upfront by entering their UPI PIN, and merchants can later debit exact amounts as services are used. This facility is most commonly used for subscription based services such as Netflix payment every month.
Pine Labs says these existing payment rails can now be adapted for AI.
According to the company, “The consumer authorises once, upfront. After that, the agent browses, selects, negotiates, and pays. No human authentication. No interruption. No friction.”
How does the system keep AI under control?
One obvious question is security. If AI can spend money, who makes sure it does not overspend? Pine Labs says this is where Grantex comes in. Grantex is a digital identity and permission layer for AI agents. It allows AI systems to prove who they are, receive permission to spend, operate within approved spending limits, and maintain records of every transaction.
The system also uses HTTP 402, which acts like a common digital language that lets systems request and process payments in a standardised way.
Together, these technologies are designed to allow AI systems to buy, sell, and make payments on behalf of businesses without requiring human approval at every step.
It is already being used
According to Pine Labs, the system is already live on Gullak. For example, a Gullak user can set a simple instruction such as: “Buy Rs 500 worth of gold if the price drops below Rs 16,000 per gram,” and approve a UPI mandate once. After that, Gullak’s AI agent keeps tracking gold prices automatically and completes the purchase when the condition is met. Instead of asking the user to approve the payment again, the app simply sends a confirmation once the transaction is completed.
Could this change online shopping?
Pine Labs says the same idea can extend to e-commerce. Karan Gupta, Managing Director, Vijay Sales, explained the potential use case. “P3P fundamentally alters this dynamic. Instead of a customer constantly checking back for discounts on a smartphone or refrigerator, they can now deploy an AI agent to secure that product the moment it hits their target price. It’s like giving every customer a personal shopper who never misses a deal,” he says.
The bigger question: Rules still do not exist
While the technology may be ready, regulation remains unclear. There is currently no RBI framework that defines how AI agents should operate on India’s UPI payment system. That creates a regulatory gap.
UPI One Time Mandates and Reserve Pay were originally built for specific purposes — enabling smoother access to services like IPO applications and recurring subscription payments. Using those systems to allow autonomous AI agents to make purchases introduces a new layer of questions around responsibility, limits, consumer protection, and oversight.
– Ends
Source link
[ad_3]
