Razorpay’s AI-powered payments foundation model, trained on four billion transactions, promises to boost success rates, enhance fraud detection, and personalise checkouts in India’s expanding digital economy, signalling a significant technological leap.
Razorpay says it has built an artificial intelligence payments foundation model trained on four billion transactions, a move it says could lift payment success rates, sharpen fraud detection and make checkouts more personalised as India’s digital payments market continues to expand. The Hindu BusinessLine reported that the system was developed with NVIDIA and AWS and is designed to analyse payment flows across merchants, banks, payment instruments and gateways.
The company said early testing across 1.5 million transactions and more than 51,000 businesses showed an 8% to 10% improvement in payment success rates, an eightfold increase in detection of international card fraud and a fivefold rise in identifying fraudulent or disputed transactions. Harshil Mathur, Razorpay’s co-founder and chief executive, said the model has already been adopted by some merchants for a significant share of their payment volumes, with Blinkit seeing a one to two percentage point lift in success rates.
Razorpay is pitching the model as different from conventional machine-learning tools, which are usually built for one task at a time, such as fraud screening or payment routing. Mathur said the company believes India needs its own approach because the country’s payments stack, including UPI, cards and net banking, is more complex than systems in many other markets.
The company said the model is being operated within India and that personally identifiable information is removed before data is fed into the system. Razorpay also said it has kept guardrails in place to avoid worsening false positives in fraud detection. For now, the company is not charging merchants separately for the technology, and Mathur said it expects to monetise the system indirectly through higher payment volumes and related services, including possible use cases in lending and marketing.
The new model fits into a broader AI push by Razorpay. Company announcements show it has also integrated its payment gateway into OpenAI’s Codex to let developers add payments to applications more quickly, and it has worked with the National Payments Corporation of India and OpenAI on “agentic payments” in ChatGPT, where users can discover and pay for products through conversational interfaces. Razorpay has also linked its existing routing tools and success-rate analytics to the same wider aim: using data-driven automation to improve how money moves across India’s digital economy.
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.
