Ecommerce platform Meesho has claimed that more than 75 per cent of orders on its marketplace now originate from AI-powered personalised feeds, signalling a growing shift towards discovery-led shopping among Indian consumers.
The company on Thursday disclosed details of PRISM (Personalised Ranking and Intent Signal Module), its in-house artificial intelligence system that powers product recommendations, search rankings and trend analysis across the platform.
The disclosure offers a glimpse into how ecommerce companies are increasingly relying on recommendation engines to influence purchase decisions, particularly among consumers who browse products rather than search with a specific intent.
According to Meesho, PRISM supports a platform that served 264 million annual transacting users and processed 717 million orders in the fourth quarter of FY26. The company said the system analyses behavioural, transactional and contextual signals in real time to personalise product discovery.
Beyond Search-led Commerce
For years, online marketplaces were built around search bars, assuming consumers knew exactly what they wanted to buy. Meesho argues that this model is becoming less relevant for a large section of India’s internet users, particularly those in smaller cities and towns.
“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,” said Debdoot Mukherjee, Chief Data Scientist and Head of AI and Demand Engineering, Meesho.
The company said PRISM supports shopping experiences in more than 10 Indian languages, including Hindi, Bengali, Marathi, Tamil, Telugu, Kannada, Malayalam, Gujarati, Punjabi and Odia.
Meesho also said the system is powered by more than 100 AI ranking models and runs on BharatMLStack, its in-house machine-learning infrastructure platform.
Focus On Discovery
The company claims that PRISM continuously evaluates user behaviour and shopping journeys to surface products that are most likely to generate engagement and purchases. Through its Trendpulse engine, Meesho said the platform identifies emerging demand patterns across regions and consumer segments.
“PRISM brings together real-time intelligence and ranking systems to understand how Bharat discovers products, expresses intent and makes purchase decisions online,” Mukherjee said.
The company further claimed that the system is trained on more than 400 trillion signals and executes over six trillion inferences daily. During peak sale events, PRISM reportedly processes nearly 100 million inferences per second. These figures could not be independently verified.
The company said that this comes as ecommerce companies race to strengthen their AI capabilities amid rising competition from quick-commerce and social-commerce platforms. Industry observers say recommendation engines are increasingly becoming the primary interface through which consumers discover products online, mirroring trends seen across social media and content platforms.
While Meesho’s disclosures highlight the growing scale of AI deployment in ecommerce, the commercial impact of such systems will ultimately be measured by their ability to improve customer retention, spending patterns and seller outcomes.
As India’s online retail market expands beyond metropolitan centres, the battle for consumers may increasingly be determined by which platform can predict what shoppers want before they actively search for it.
