Razorpay has partnered with Sarvam to integrate conversational AI with payments infrastructure, enabling voice-first commerce across Indian languages.
The collaboration combines Sarvam's multilingual AI models and agentic stack with Razorpay's payments infrastructure, allowing users to discover products, place orders, and complete payments through spoken conversation in their preferred language.
The initial deployment is being piloted through Sarvam's chat platform, Indus App, with food delivery platform Swiggy as the first merchant partner. Within that environment, customers will be able to order food by speaking to an AI assistant, which manages the full transaction, from interpreting the request and placing the order to completing payment via the National Payments Corporation of India's UPI Reserve Pay. The entire process is set to take place within a single conversation.
Addressing a structural gap in Indian digital commerce
The partnership is framed around a specific market dynamic. India has over 950 million internet users and more than 450 million UPI-registered digital payment users, according to figures cited from the Internet in India Report 2025, published by IAMAI and Kantar, and data from NPCI. Yet only approximately 200 million people currently shop online. The gap is attributed in part to the English-language orientation and navigational complexity of conventional ecommerce applications, which present barriers for a significant portion of the digitally connected population.
With this in mind, the voice interaction in regional languages is positioned as a way to close that gap, moving commerce away from app-based interfaces and towards natural conversation as the primary channel.
Three layers of deployment
The integration is structured across three areas. First, conversational commerce within Indus App, where the Swiggy pilot is already under development. Second, an embedded voice commerce capability for third-party websites and applications, as an early version has been built for The Derma Co's website, enabling product discovery and purchase by voice. Third, Sarvam's AI stack will be integrated into Razorpay's Agent Studio, giving developers the tools to build multilingual conversational agents that support Hindi, Hinglish, and other Indian languages within the Razorpay ecosystem.
The arrangement reflects a broader shift in how payments infrastructure providers are positioning themselves relative to AI-driven interfaces. By embedding payment execution directly within conversational agents, the model removes the redirect to a separate checkout flow that characterises most current ecommerce transactions.
