Imagine you’re asking an artificial intelligence (AI) assistant the seemingly simple question, “What to order for dinner”, and during that same conversation it seamlessly progresses to the “order is on the way” stage. This is the world of agentic AI payments envisioned by Razorpay, India’s largest payments aggregator. At the India AI Impact Summit 2026, the fintech has announced agentic payments on Anthropic Cloud. As Razorpay chief operating officer Rahul Kothari told HT, “commerce is being reengineered”.
It is Razorpay’s Agent Payments feature, the coming together of India’s UPI infrastructure and cloud conversational intelligence led by the National Payments Corporation of India (NPCI), that hopes to transform everyday AI interactions into real, complete purchases. Additionally, Razorpay is also partnering with global agentic software creation platform Replit to help AI developers integrate monetization options into the apps they build.
The Anthropic AI-led payments chapter comes just months after Razorpay launched agentic payments on OpenAI’s ChatGPT as well as connectors for Google Gemini. Kothari believes that trust design will matter more in the new era of AI commerce. “We are moving towards a world where checkout will become invisible and so trust in payments will be critical,” he says.
Crucial to agentive payment aspirations is UPI Reserve Pay, a feature that allows users to reserve a specified amount from their RuPay credit card, a bank account or pre-approved credit line on UPI and get paid from that pool. Razorpay and Anthropic’s agentic commerce chapter currently have Zomato, Swiggy and Zepto as partner platforms.
“UPI is really the secret weapon for agented commerce as well. It lets you block funds for an agent without sharing payment credentials. No other real-time payment or card system does this natively,” says Kothari. With over 55% share, Razorpay is the dominant player in India’s online payment gateway sector.
When asked more general questions of the AI chatbot such as “order dinner for people under two.” ₹800 from my favorite restaurant”, will be able to find options on Zomato and Swiggy, and complete the payment after user confirmation. Kothari emphasizes the main idea behind integrating agentive payments to work with popular instant commerce and food delivery apps is to redefine how intention becomes an economic action, in an envelope of familiarity.
Regulatory conversations are inevitable
Current payment systems are designed on a fundamental assumption that a human has made the decision to make the payment. Liability frameworks, dispute mechanisms and fraud prevention are all based on this assumption. Agent payment breaks it down completely.
Kothari points out that over the past months, his learnings about agentic payments trends point to a pleasing trend – users are actually not afraid to make AI payments. But he draws a line in the sand. “They fear losing control, and the challenge for us is to solve that with real-time tracking, instant revocation, and user-defined spending limits,” he says.
When an AI books a flight for you and the airline charges twice, who is liable – you, the AI platform, the payment processor, or the airline? When there is no checkout page, consent is legally undefined. When an AI misinterprets a user’s instructions and buys the wrong thing, is that a fraud dispute, a product dispute, or something that existing rulebooks don’t currently address?
Regulators will have to answer this question at some level by defining who is liable when AI makes an error in a transaction?
“India is best placed to lead here because the Reserve Bank of India and NPCI have always been innovation-friendly. There should be guardrails. UPI was a regulatory innovation, and the same methodology is applied in AI commerce,” says Kothari, before pointing out the complexity that “AI will keep evolving and you can’t really write rules for every AI agent”.
He hopes, “Regulators in India will take the lead on how governance should be done,” suggesting that mechanisms such as user-based consent for payments above a certain value per transaction, ability to revoke permission to make payments and shift of liability should be phase-based.
For those unaware, India’s digital payments trajectory needs no example. Official data from NPCI shows that UPI transactions have been completed ₹₹28.33 lakh crore in value and ₹21.7 billion in volume in January – the highest monthly figures for UPI. Also, spend on credit cards in India ₹₹2.12 lakh crore for 552 million transactions, just a little ₹A record of Rs 2.17 lakh crore made in September.
For agent payments to be successful, in terms of adoption, Kothari believes that the payments infrastructure needs to be platform agnostic. “The AI layer, whether it’s ChatGate or Gemini or the cloud, is just the interface. The payment system has to work the same way, even if there’s AI on top,” he says. The word he uses is “non-negotiable”.
Razorpay has maintained significant momentum over the past few months, including receiving an Offline Payment Aggregator (PA-P) license from the RBI to collect in-store payments last month.
