Bengaluru-based voice AI startup Gnani.ai has partnered with fintech firm Razorpay to roll out a new AI-powered collections platform that can complete payment transactions during live customer calls.
The companies say this is the first time an automated voice system can not only remind customers about pending dues but also close the payment within the same conversation.
For many enterprises, automated collections have always had a limitation. Systems could place reminder calls, but once a customer agreed to pay, the process shifted elsewhere — to a link, an app, or a human agent. That break often meant lost conversions.
This new platform changes that flow. Gnani.ai’s enterprise voice system, which handles over 10 million calls a day, is now integrated with Razorpay’s Model Context Protocol (MCP) Server — an AI-native payments layer that allows transactions to be initiated and completed in real time.
Closing the Loop: From Payment Reminder to Real-Time Settlement
During a call, the AI agent can generate a payment link, trigger a UPI collect request, monitor the payment status, and confirm settlement before ending the conversation. The system also evaluates customer intent and suggests suitable payment options automatically.
Ganesh Gopalan, Founder and CEO of Gnani.ai, said the integration moves voice automation beyond simple conversations. Instead of just following up on dues, enterprises can now potentially complete collections at scale without manual intervention in routine cases.
At present, the platform supports standalone UPI payments for one-time transactions. It also enables UPI Single-Block-and-Multiple-Debits (SBMD), or Reserve Pay, which is used for mandate-based recurring payments. Card and wallet options are expected to be added later.
Khilan Haria, Chief Product Officer at Razorpay, noted that AI becomes far more useful when it can take action. By allowing AI systems to securely execute payments inside live conversations, collections teams can move from repeated reminders to confirmed payments in a single interaction.
Razorpay’s MCP Server functions as a secure bridge between the AI agent and the payments infrastructure. It enables AI-driven commands such as generating links, checking payment status, and confirming settlement — without relying on traditional dashboards.
According to Gnani.ai, its proprietary automatic speech recognition system improves accuracy by about 30% compared to competing technologies. The platform supports conversations in more than 38 languages, including Hindi, Bengali, Punjabi, Tamil, Telugu, Marathi, Gujarati, and Kannada.
The process is straightforward. The AI verifies outstanding dues, confirms payment intent, generates a transaction-specific link, and sends it via SMS or WhatsApp. It continues tracking the payment in real time through system integrations during the same call.
The companies clarified that sensitive payment details are handled entirely within Razorpay’s infrastructure and do not pass through the voice AI system. Links are generated with unique transaction parameters to prevent reuse, and UPI mandate-based payments require customer approval through registered UPI apps.
Gnani.ai also stated that its systems are GDPR compliant and SOC 2 Type II certified.
Separately, Razorpay has been working with the National Payments Corporation of India on an agentic payments pilot built on Anthropic’s Claude model. The initiative allows users to complete purchases — such as ordering food from Zomato or groceries from Swiggy and Zepto — directly within a conversational AI interface without switching apps.
Earlier this year at the India AI Impact Summit 2026, Gnani.ai introduced two new foundational voice AI models under its Inya VoiceOS stack, including a five-billion-parameter voice-to-voice system, as part of India’s broader push to strengthen domestic AI capabilities.
Also Read: Razorpay Partners with Replit to Enable UPI Payments for AI-Built Apps in India
