ThirdWatch, Stopping Fraud and RTO using AI

ThirdWatch operates as a data science subsidiary of Razorpay, providing real-time, automated fraud prevention solutions tailored for the e-commerce sector. Established in 2016 by co-founders Adarsh Jain and Shashank Agarwal, the Gurugram-based company was acquired by Razorpay in August 2019 to bolster Razorpay's capabilities in combating online transaction fraud. The founders' journey began from a shared frustration with e-commerce limitations caused by blunt anti-fraud measures, such as blocking entire pin codes, which negatively impacted genuine customers. This personal experience, coupled with their background in building a mobile analytics system, spurred the creation of ThirdWatch. Adarsh Jain, an alumnus of IIT-BHU, and Shashank Agarwal, an ethical hacker and programmer, combined their expertise in big data and technology to address these market gaps.

The company's business model is centered on a Software-as-a-Service (SaaS) offering that helps e-commerce merchants mitigate risks and reduce losses from fraudulent activities. ThirdWatch's core product, an AI-driven platform named 'Mitra', is designed to prevent various types of fraud, including those related to payments, promotional code abuse, and significantly, Return to Origin (RTO) issues, a prevalent problem in India's cash-on-delivery-heavy market. By integrating with e-commerce platforms like Shopify, the service acts as a simple plug-in, making it accessible to a wide range of online businesses. Revenue is generated by providing these advanced fraud detection tools to merchants, which after the acquisition, became available to Razorpay's extensive client base.

The Mitra platform analyzes over 200 parameters for each transaction in real time to generate a trust score, flagging it as either legitimate or suspicious. It leverages machine learning algorithms and a variety of data points including device fingerprinting, location profiles, user behavior analysis, IP address verification, and account profiling. A key feature is its ability to learn from transaction data across its network, creating a network effect that enhances the algorithm's accuracy over time. This allows the system to identify subtle, complex patterns indicative of fraud that simple rule-based systems often miss, thereby reducing false positives and helping businesses increase their operational profitability.

Keywords: fraud prevention, e-commerce security, artificial intelligence, machine learning, real-time fraud detection, Return to Origin, RTO reduction, payment fraud, SaaS, fintech, Razorpay, Adarsh Jain, Shashank Agarwal, Mitra AI, device fingerprinting, transaction monitoring, risk scoring, cash on delivery fraud, online merchant tools, cybersecurity