Razorpay says its WhatsApp-based AI support agent has reduced customer service volume by 30% to 40%, with voice interactions accounting for about 60% of use.
Razorpay says its WhatsApp-based AI support agent has cut the company’s customer service volume by nearly 30% to 40%, as merchants increasingly choose to resolve routine payment and technical issues through conversational tools rather than conventional support channels. Harshil Mathur, the company’s co-founder and chief executive, said at the ElevenLabs Summit in Bengaluru that voice has become a surprisingly important part of the experience, accounting for about 60% of interactions with the bot.
The shift reflects a broader push by Razorpay to move merchant support, payment operations and business workflows into agentic AI systems that can do more than answer simple questions. The company has been rolling out tools through its Agent Studio platform, which is designed to let businesses automate tasks such as subscription recovery, payment operations and customer engagement while still keeping merchants in control of how the systems behave.
Razorpay has also been broadening its conversational products beyond support. In September, it introduced RAY, an AI account manager on WhatsApp built with IndusInd Bank, allowing businesses to check payments, settlements and support queries by text or voice notes. The company said at the time that RAY had already handled more than 20,000 conversations, while earlier this year it also launched agentic payments for voice AI, a system intended to move voice interactions closer to actual transaction execution.
At the same time, Razorpay’s expansion into AI agents comes against a more cautious backdrop for the wider industry. A survey cited by ITPro found that many companies have rolled back or shut down customer service AI agents because of concerns about data exposure, hallucinations and weak auditability. Razorpay’s approach appears to be aimed at addressing some of those issues through controlled workflows, consent rules and human review, even as the company says its own engineers are already using AI to diagnose ticket trends and draft code fixes for review.
Disclaimer: This article is intended to inform and educate, not to recommend or endorse any financial product, investment or strategy. Please consider your own financial circumstances and seek professional advice where appropriate before making financial decisions.
