Razorpay has launched Vulcan, a transformer-based AI foundation model built specifically for digital payments, as the fintech company looks to make online transactions more reliable, secure and efficient.

Developed with technology from NVIDIA and AWS, Vulcan is designed to analyse payment behaviour and make real-time decisions across different stages of a transaction.

Unlike conventional large language models that are designed to understand and generate text, Vulcan has been developed to understand payment data and transaction patterns. Razorpay says the model’s architecture and training data are proprietary, allowing it to learn from the movement of money across its payments network.

The model has been trained using nearly 3 trillion data points from more than 4 billion payments. It analyses around 3,000 signals for each transaction, allowing it to identify patterns that can be used for payment routing, fraud detection, risk assessment and checkout optimisation.

Razorpay says early deployment of components of Vulcan has produced significant improvements. Payment success rates have increased by around 8% to 10%, while the system has detected up to eight times more international card fraud. It also identified five times as many fraudulent or disputed transactions without increasing the number of alerts.

The company developed Vulcan after studying payment behaviour involving around 1.5 million shoppers and more than 51,000 businesses. The research highlighted payment-related friction across both major cities and smaller markets, reinforcing the need for systems that can make transactions more predictable.

Another focus is payment routing. Vulcan can evaluate different payment routes in real time and identify the option that is more likely to result in a successful transaction. This could be particularly important in India’s highly diverse payments ecosystem, where consumers use multiple banks, cards, UPI applications and payment methods.

Razorpay plans to expand Vulcan beyond its current applications into areas such as lending and payment authentication. This could allow the company to use the same AI infrastructure across a wider range of financial services.

The launch also highlights the growing role of AI in India’s fintech industry. Rather than using AI only for customer service or automation, companies are increasingly applying it to the underlying infrastructure that determines whether transactions succeed, whether fraud is detected and how financial risk is assessed.

For Razorpay, Vulcan could become an important technology layer as India’s digital payments market continues to expand. The company’s ability to improve transaction success while controlling fraud and maintaining security will be key to determining the model’s long-term impact.

Feeling overwhelmed by the markets? Let Tradz by EquityPandit be your guide. Our user-friendly app simplifies complex data and provides actionable trading signals. Download the app today and trade with confidence!