Razorpay Vulcan
Why in News?
Razorpay has launched Vulcan, India’s first transformer-based AI foundation model built specifically for digital payments. The system went live on 18 August 2026.
Technology & Training Scale
- Massive Dataset: Vulcan is trained on nearly 3 trillion data points across 4 billion transactions.
- Deep Analysis: The system evaluates approximately 3,000 signals per transaction.
- Real-time Processing: It completes its entire analysis and data routing decision in just 29 milliseconds before a transaction goes through.
- The "Language of Money": Unlike Large Language Models (LLMs) that understand text, Vulcan uses transformer technology to understand the flow and relationships of money movement.
The Core Shift: Unified vs. Fragmented AI
- Single Intelligence Layer: Traditional setups use separate machine learning models for routing, fraud, risk, and checkouts.
- Shared Context: Vulcan replaces these fragmented tools with one cohesive layer that handles all parameters simultaneously.
- Compounding Intelligence: The system is self-improving, meaning every processed payment teaches it how to optimize the next transaction.
Real-World Early Impact Metrics
During its beta testing across more than 51,000 businesses (including brands like Blinkit, Bachatt, and redBus), Vulcan delivered the following metrics:
During its beta testing across more than 51,000 businesses (including brands like Blinkit, Bachatt, and redBus), Vulcan delivered the following metrics:
- 8% to 10% improvement in payment success rates.
- 8x increase in detecting and stopping international card fraud.
- 5x more fraud or disputed transactions identified without increasing false alerts for businesses.
- 40% more shoppers are successfully shown their preferred UPI app on Razorpay Magic Checkout.
- 1 to 2 lakh additional purchases are completed every single month due to smart checkout personalization.
- Zero Cost for Merchants: Razorpay is providing this advanced AI layer as a standard infrastructure upgrade without charging merchants extra fees.
