Razorpay has launched Vulcan, an AI model built specifically for digital payments.
So, what does it actually do: before you pay, Vulcan can pick the payment route most likely to work, flag suspicious transactions and predict which payment method you’re most likely to successfully use.
It’s powered by NVIDIA GPUs and AWS infrastructure, and parts of the technology are already being used in live payment environments.
These include companies like Blinkit, Bachatt and redBus. Unlike ChatGPT, it isn’t trained on language, it’s trained to understand patterns in payments.
Why does this matter: India’s digital payment boom has also meant more fraud. In FY25, 12.6 lakh UPI fraud cases involving ₹981 crore were reported. At the same time, UPI processed a massive 20.47 billion transactions worth ₹26.32 lakh crore in November alone.
Razorpay says Vulcan has improved payment success rates by 8-10%, detected 8x more international card fraud, and identified 5x more fraudulent or disputed transactions without increasing alerts. It can make these decisions in just 29 milliseconds, significantly making digital payments safer and more reliable.


