Can Razorpay Turn ChatGPT Into India’s Next Commerce

What happened

ChatGPT is beginning to emerge as more than a search assistant for ecommerce. For a growing number of brands and marketplaces, it is also becoming a new channel for product discovery and, increasingly, shopping. Nykaa offered one of the clearest signals of that shift in its recent Q1 FY27 management commentary.

The beauty retailer said it has become the top beauty brand on ChatGPT and Gemini through what it calls “Answer Engine Optimization” (AEO), and indicated that AI answer engines are emerging as a customer acquisition channel alongside Google Search. It also announced plans to bring Nykaa Beauty and Nykaa Fashion shopping experiences into ChatGPT as part of a broader OpenAI partnership.

Months before AI commerce entered mainstream investor conversations, the payments company began working with OpenAI and NPCI on infrastructure that enables merchants to build ChatGPT-native storefronts and complete purchases within the chat interface.

The experience will be enabled on UPI Reserve Pay, which allows users to block a specific amount of funds for future debits to designated merchants, and UPI Circle, a solution that delegates UPI authentication, enabling payments to be completed directly within ChatGPT without switching to external apps or websites.

Razorpay sees its role extend beyond payments to helping merchants integrate product catalogues and storefronts into AI platforms, while using NPCI’s UPI Circle and UPI Reserve Pay to enable native checkout. Rival Cashfree, meanwhile, has launched “Cashfree Here”, a payments extension for AI applications that enables in-chat transactions on platforms such as ChatGPT and Claude, highlighting how payment companies are taking different approaches to AI-native commerce.

Mathur, while disclosing the AI roadmap in an earlier interview with us, had stated that the real fight in Indian commerce over the next few years won’t be over who owns the checkout button, but over who owns the conversation that precedes it — and that whoever wins that layer effectively controls the transaction.

Besides ChatGPT, Razorpay is also looking at integrations with other AI models like Anthropic’s Claude and Google’s Gemini for agentic payments products. Razorpay has framed the shift as setting up shop inside ChatGPT or ChatGPT Storefronts that would otherwise require a merchant to host their own MCP server and build a new checkout flow from scratch — commonly an eight-to-ten-week engineering effort.

The pitch is that startup collapses this timeframe into something closer to half an hour for a Shopify merchant, auto-syncing the catalogue and handling the submission process to OpenAI on the brand’s behalf. Cashfree, on the other hand, has got started with agentic payments products, but that will still require the merchants to build their own AI agents which could be plugged into Cashfree’s MCP layer.

“The exclusive part is on the merchant side,” Mathur told Inc42. ” He said AI platforms will continue to support multiple payment providers, much like a merchant can host a website on Google while connecting any payment gateway. However, merchants are likely to work with only one or two payment providers for agentic commerce, with Razorpay aiming to secure those relationships early.

Mathur said Razorpay currently has the largest number of merchants live on these platforms, although the company declined to disclose updated numbers, saying adoption is scaling rapidly. Trust, however, remains a common concern across the industry.

Cashfree’s CEO Akash Sinha earlier told Inc42 that although AI holds a huge promise for bringing efficiencies in the payments infrastructure business, the agentic commerce layer will likely face the challenge of consumers trusting agents with payments and data. Mathur argues that Razorpay’s underlying payment architecture has been designed to ensure users remain in control of every transaction.

“The way we have built our integration is we are using the Reserve Pay protocol which ensures that the payment data doesn’t touch the LLMs at any point of time,” Mathur said. According to Mathur, payments must either be backed by pre-consent from the customer or require two-factor authentication.

“Both of those are mandated,” he said, adding that ensuring compliance and governance while enabling payments through AI platforms is part of Razorpay’s role as the payment infrastructure provider. The broader regulatory framework around AI-led commerce, however, is still evolving as more participants enter the transaction flow The early adopters are unsurprisingly digital-first brands.

Why it matters

According to him, AI models such as ChatGPT do not receive customers’ payment credentials, and every transaction continues to require user authorisation through NPCI’s UPI infrastructure. In practice, the shopping flow begins with a user discovering products through an AI assistant. Once a purchase is initiated, the checkout is handed over to Razorpay’s payment infrastructure, which uses NPCI’s UPI capabilities—including UPI Circle and UPI Reserve Pay—to complete the transaction.

What to watch

While the shopping experience remains conversational, payment authorisation still occurs through the user’s own UPI account and applicable two-factor authentication requirements, rather than through the AI assistant itself. “It’s not like you give a customer’s details and let the agent shop and the customer has no control,” Mathur told Inc42.