Razorpay builds proprietary AI model Vulcan to boost payment success rates

Trained across four billion transactions and 3 trillion data points, Vulcan aims to improve payment success rates, detect fraud and personalise the checkout experience.

Payments technology company Razorpay on Tuesday rolled out an artificial intelligence (AI) model Vulcan trained across four billion transactions to improve payment success rates and detect frauds. 

 

The Bengaluru-based company said the model was proprietary to the company with the architecture and training data belonging to the fintech firm. 

 

It claims to have been trained across 3 trillion data points with the model’s ability to understand nearly 3,000 signals from each transaction. What is different in this model is that instead of a specialised model each for routing, risk, fraud and checkout, a single model like this understands all data points at one point.

   

“An AI-led payments foundation model doesn't just solve today's problem and stop there. Every payment teaches the system something that makes the next payment better,” said Harshil Mathur, chief executive officer (CEO) and founder, Razorpay. 

 

The model, built with expertise from Nvidia and AWS (Amazon Web Services), enables routing, detection of frauds at the network level, flags risky cash on delivery transactions, and allows for predictive checkout personalisation. 

 

Razorpay’s said that its study across 1.5 million shoppers and more than 51,000 businesses found the same payment friction - failed transactions, drop-offs, delays - surfacing identically from a metro high street to a small-town market. 

 

It said the model had been capable of an improvement of 8 to 10 per cent in transaction payment success rates. It saw 40 per cent more shoppers spot their preferred UPI app on checkout helping to complete 1 to 2 lakh more purchases each month. 

 

“8x more international card fraud detected and stopped… 5x more fraudulent or disputed transactions identified, without increasing the number of alerts…” the firm said.