Indian payments firm Razorpay unveils an advanced artificial intelligence model trained on billions of transactions, significantly enhancing payment success rates and fraud detection capabilities while paving the way for AI-driven financial services.
Razorpay has built an artificial intelligence payment foundation model trained on 4 billion transactions, in a bid to lift payment success rates, sharpen fraud detection and tailor the checkout experience, according to the company. The Indian payments firm said the system studies more than 3,000 signals in each transaction, mapping how money moves between merchants, banks, payment instruments and gateways.
The company said early tests on 1.5 million transactions and more than 51,000 businesses showed payment success rates rising by as much as 8% to 10%, while detection of suspicious or disputed transactions improved fivefold. Razorpay said international card fraud detection was also boosted eightfold in those trials. The model was developed with NVIDIA and AWS.
Harshil Mathur, Razorpay’s co-founder and chief executive, said the appeal of the model is that it can handle a wide range of payment tasks through one underlying system, rather than using separate models for each job. He said some merchants have already routed 50% to 60% of their payment volume through the technology, and that Blinkit saw a 1 to 2 percentage point improvement in payment success rates.
Razorpay is positioning the model as part of a broader push to link financial services more closely with AI. The company said the system is built and operated in India, with personally identifiable information removed before data is used. It also said it has kept fraud false-positive rates stable while expanding the system’s reach into authentication, routing, fraud and lending. Mathur said Razorpay does not expect to charge merchants directly for the technology in the near to medium term, and instead plans to monetise it through wider services and more payment processing.
The new model builds on Razorpay’s earlier AI efforts, including Optimizer, a routing system the company said was designed to cut payment failures that can cost businesses more than ₹7,000 crore a year. Razorpay has also been working with the National Payments Corporation of India and OpenAI on agentic payments, a pilot aimed at enabling AI-driven shopping and checkout, and it recently launched an MCP Server for instant AI payment integration.
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.
