Meesho Limited has announced the significant impact of its proprietary AI intelligence system, PRISM (Personalised Ranking & Intent Signal Module). This system is designed to revolutionize how 264 million Indians discover products and shop online. PRISM is built to cater to a unique shopping behavior prevalent in India, where consumers often browse and discover products rather than searching with specific intent. For the platform's 264 million annual transacting users and 717 million placed orders in Q4 FY26, PRISM shifts commerce from keyword-led search to real-time intent understanding. Currently, over 75% of orders on Meesho originate from AI-driven personalized feeds powered by PRISM. The system's architecture interprets behavioral, transactional, and contextual signals across users, utilizing over 100 AI ranking models trained on more than 400 trillion input signals, executing over 6 trillion inferences daily. It supports 10+ multilingual experiences and is powered by Meesho's in-house ML infrastructure platform, BharatMLStack. Debdoot Mukherjee, Chief Data Scientist, Head of AI and Demand Engineering at Meesho, stated that the next wave of Indian internet users will discover rather than search, and expect technology to meet them where they are. PRISM is designed to meet this evolving reality, continuously learning and anticipating user needs to make discovery effortless. During peak sale events, PRISM processes nearly 100 million inferences per second. PRISM currently enhances personalized recommendations, intent-led discovery, trend intelligence, catalogue understanding, and seller growth systems across Meesho's ecosystem, which includes 263 million monthly active users and 17 billion product views per day.
