Razorpay launched an artificial intelligence foundation model trained on nearly 3 trillion data points to reduce failed transactions and strengthen fraud detection, Fintech News Singapore reports.

The model, named Vulcan, analyses about 3,000 signals per transaction and was built using NVIDIA graphics processing units and Amazon Web Services, according to the publication. Early components running on the network have improved payment success rates by 8% to 10%.

"Every payment teaches the system something that makes the next payment better," Razorpay CEO and founder Harshil Mathur said.