Razorpay has launched Vulcan, a transformer-based AI foundation model designed to expedite digital payments by making them more reliable and secure. The model, built with NVIDIA and AWS technology, is proprietary and trained on approximately 3 trillion data points from 4 billion payments. Razorpay claims Vulcan can score payment routes in real-time and flag fraud across multiple merchants, enhancing payment success and fraud detection.
Vulcan combines Razorpay’s payments data with NVIDIA’s accelerated computing and AWS’s cloud infrastructure. Unlike traditional large language models, Vulcan learns the movement of money by analyzing around 3,000 signals per transaction across the entire payments ecosystem. Razorpay reports early deployment results showing an 8-10% improvement in payment success rates, an eightfold increase in international card fraud detection, and a fivefold rise in identifying fraudulent or disputed transactions without increasing alert volume.
The model’s ability to continuously learn from every transaction processed sets it apart from narrow problem-solving AI systems. By understanding the complex financial journey at scale, Vulcan addresses challenges in digital payments, a sector where fraud and transaction failures remain significant issues. Razorpay’s approach reflects growing industry efforts to integrate AI for enhanced security and efficiency in fintech, leveraging partnerships with technology leaders like NVIDIA and AWS.
Razorpay’s Vulcan model has already demonstrated a 40% increase in shopper engagement during early live transaction deployments. The startup, which is preparing for an IPO, positions Vulcan as a key innovation to strengthen its digital payments platform and reduce fraud risks across its merchant network.