E-commerce platform Meesho has announced the integration of PRISM (Personalised Ranking & Intent Signal Module), an internal AI system that powers product recommendations for its 264 million annual transacting users, helping millions of Indians discover products, shop online, and engage across the platform.
The company, in a release noted that unlike traditional e-commerce platforms that primarily rely on keyword-based search, PRISM analyses real-time user behaviour to predict shopping intent and deliver personalised product recommendations.
It added that over 75% of orders on Meesho originate from AI-driven personalised feeds powered by PRISM.
PRISM is a real-time intelligence and ranking architecture that uses over 100 AI ranking models trained on 400 trillion input signals. It executes 6 trillion daily inferences, which increases to 100 million inferences per second during peak traffic periods.
To process these workloads, Meesho developed BharatMLStack, an in-house machine learning infrastructure designed to reduce data processing costs compared to standard cloud options.
Meesho further noted that PRISM also includes an LLM component called Trendpulse, which tracks regional purchase patterns to display inventory matching local demand. The system supports 10 languages, focusing on voice-led navigation and regional dialects.
Speaking on the impact of PRISM, Debdoot Mukherjee, chief data scientist, head of AI and demand engineering, Meesho, stated that “The next hundred million Indians coming online will not search, they will discover. They will not type, they will speak, browse, and expect technology to meet them where they are. It supports 10 plus multilingual experiences across Hindi, Bengali, Marathi, Tamil, Telugu, Kannada, Malayalam, Gujarati, Punjabi, and Odia”.
The system currently manages product recommendations, trend analysis, and seller distribution metrics. It operates across a dataset generated by 263 million monthly active users, 17 billion daily product views, and over 2 billion combined user ratings and reviews.
