Razorpay builds AI payments model trained on 4 billion transactions
Razorpay has built a new AI-powered payments model designed to make your online transactions smoother and safer.
Built with help from NVIDIA and AWS, this tech is trained on data from 4 billion transactions and checks thousands of signals for every payment.
It's built to track how money moves across merchants, banks, and gateways, so fewer failed payments and less hassle at checkout.
Razorpay model boosted success rates 8-10%
In early tests with 1.5 million transactions, the model boosted success rates by 8% to 10% and spotted international card fraud eight times better than before.
It's tailored for India's mix of UPI, net banking, and cards, because Harshil Mathur, Co-founder and CEO of Razorpay, explained that payment foundation models built in other markets cannot simply be deployed in India because the country's payment architecture, including UPI, cards, net banking and its diverse banking ecosystem, is different.
Right now, merchants are not being charged separately for the model, but Razorpay plans to use the tech for more services like lending and marketing down the line.