As India’s retail credit market expands and lenders sharpen their focus on recovery efficiency, Bengaluru-based AI platform Gnani.ai has partnered with fintech major Razorpay to introduce an AI-driven collections platform that can complete payment transactions during live customer calls.

The companies say the integration marks a shift from conventional voice bots that merely remind borrowers about dues to systems capable of executing transactions in real time. By embedding payments infrastructure directly into the conversational workflow, the platform aims to reduce friction and improve conversion rates in collections.

At the heart of the offering is the integration of Gnani.ai’s enterprise agentic AI platform—processing over 10 million calls daily—with Razorpay’s Model Context Protocol (MCP) Server. The MCP layer enables AI agents to initiate, authorise, and complete transactions natively within Razorpay’s payments ecosystem, without redirecting customers through multiple interfaces.

“For the first time, enterprises can automate collections at scale without sacrificing conversion,” said Ganesh Gopalan, founder and CEO of Gnani.ai. “We’ve moved voice AI beyond reminders to actual transaction completion.”

Closing the last-mile gap in collections

Automated voice systems have become commonplace across banks and NBFCs for early-stage collections. However, while they are effective in reaching customers and generating payment intent, the actual transaction typically requires customers to switch apps, open payment links, or manually enter details—introducing drop-offs.

The new platform attempts to address this gap. Once a customer confirms intent to pay, the AI agent can generate a transaction-specific payment link or initiate a UPI Collect request to the customer’s virtual payment address. It then monitors the transaction status in real time and confirms settlement within the same call, without human intervention on the merchant side for routine cases.

Khilan Haria, chief product officer at Razorpay, said the value of agentic AI lies in its ability to act. “Collections traditionally involve multiple steps and follow-ups. By enabling AI systems to securely complete payments within a live interaction, we are unlocking a more seamless operating model for financial institutions,” he said.

The platform currently supports one-time UPI payments and UPI SBMD (also referred to as Reserve Pay) mandate-based payments for recurring obligations. Support for cards and wallets is expected to be added in subsequent phases.

Enterprise focus and compliance

Given the regulatory sensitivity around payments and customer data, both companies emphasise that transaction processing remains entirely within Razorpay’s secure infrastructure. Payment links are generated with transaction-specific parameters to prevent reuse or tampering, and UPI mandate flows require standard customer authorisation through registered apps.

Gnani.ai says its platform is GDPR compliant and SOC 2 Type II certified, aligning with enterprise-grade security and governance standards. The company’s conversational AI stack supports over 38 languages, including Hindi, Tamil, Telugu, Bengali, Marathi, and Kannada—an important capability for institutions operating across India’s diverse borrower base.

Founded in 2016, Gnani.ai serves more than 200 enterprise clients across BFSI, insurance, automotive, and government sectors. It is backed by Samsung Ventures and has raised over $7.5 million to date.

Strategic implications

The partnership reflects a broader evolution in enterprise AI—from assistive systems that provide information to autonomous agents capable of completing end-to-end workflows. For lenders grappling with rising delinquencies and cost pressures, improving “first-call resolution” in collections can materially impact recovery ratios and operational efficiency.

For Razorpay, the move extends its positioning from payments processor to infrastructure layer for AI-native commerce and financial workflows. For Gnani.ai, it strengthens its pitch as an agentic AI platform rather than a traditional voice automation provider.

As financial institutions seek higher productivity per interaction and tighter integration between communication and transaction systems, the ability to convert intent into settlement within a single conversation could mark a significant shift in how digital collections are executed at scale.