Fintech company Razorpay has launched Razorpay Vulcan, an AI-powered payments foundation model designed to make digital transactions more reliable, secure and predictable as India’s digital payments ecosystem continues to expand.
According to the company, Vulcan is India’s first transformer-based AI foundation model built specifically for payments. Developed using technology from NVIDIA and Amazon Web Services (AWS), the model is designed to address challenges including payment failures, fraud, transaction routing and checkout friction.
ndia’s digital payments landscape includes UPI, cards, net banking, wallets and cash-on-delivery transactions, with payments moving through hundreds of banks and gateways.
Razorpay said Vulcan is designed to analyse these payment routes in real time and identify the most suitable route before a transaction is attempted.
The model is intended to eventually support decisions across authentication, payment routing, fraud detection and lending.
Razorpay said it conducted an internal study involving 1.5 million shoppers and more than 51,000 businesses to understand payment-related friction across metropolitan and smaller markets.
Early components of Vulcan have also been deployed across approximately 3 trillion data points on Razorpay’s payments network for testing routing, fraud and risk decisions on live transactions.
According to the company, these deployments have produced several improvements.
Razorpay said the technology has helped increase payment success rates by up to 10 per cent and detect and prevent eight times more international card fraud.
The system has also identified five times more fraudulent or disputed transactions without increasing the number of alerts.
On Razorpay’s Magic Checkout platform, the company said 40 per cent more shoppers were able to see their preferred UPI application. This helped facilitate an additional 1–2 lakh purchases each month, according to Razorpay.
Vulcan has been developed using technologies from NVIDIA and AWS, with AWS saying the foundation model was built on Amazon SageMaker.
The technology is intended to consolidate large volumes of transaction insights into a unified AI layer rather than relying solely on multiple fragmented machine-learning models.
Razorpay CEO and Co-founder Harshil Mathur said the objective was to make digital payments more dependable, particularly for consumers who remain hesitant about choosing digital transactions over cash.
As online commerce and digital payments continue to grow, payment failures, fraud and routing decisions can directly affect both consumers and businesses.
Razorpay’s approach seeks to use AI to make these decisions dynamically, potentially allowing payment systems to respond more effectively to changing transaction conditions.
The company plans to expand Vulcan’s capabilities across additional payment functions as India’s digital commerce ecosystem develops.
