Now Meesho has more than 75% orders

Meesho said its PRISM recommendation engine is helping shift product discovery away from traditional keyword searches on the platform that serves 264 million annual transaction users.

Meesho said more than 75 percent of orders on its platform now come from AI-powered personalized product feeds powered by its proprietary recommendation engine, reflecting the growing role of artificial intelligence in shaping the way consumers discover products online.

The company said the system, known as PRISM, analyzes user behavior and intent signals in real-time to present products to shoppers who are often browsing rather than searching with a specific purchase in mind. Meesho, which reported 264 million annual transacting users and 717 million orders placed in Q4FY26, said technology is helping product discovery shift away from traditional keyword-based searches.

“The next hundred million Indians coming online won’t search, they will discover. They won’t 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 at Meesho.

Meesho said PRISM is powered by a network of more than 100 AI ranking models and trained on more than 400 trillion input signals, executing more than 6 trillion predictions within milliseconds every day. Additionally, it enables sellers to more effectively reach high-intent audiences by presenting their products to users who are most likely to engage and buy.

It supports over 10 multilingual experiences in Hindi, Bengali, Marathi, Tamil, Telugu, Kannada, Malayalam, Gujarati, Punjabi, and Oriya. The system is powered by BharatMLStack, Meesho’s in-house machine learning (ML) infrastructure platform built to support high-throughput AI workloads at significantly lower inference costs than traditional cloud infrastructure.

While traditional e-commerce platforms were designed for consumers searching with explicit intent, PRISM enables a more intuitive and search-based shopping experience for India. Through its large language model (LLM) driven search engine TrendPulse, PRISM explains emerging demand patterns across regions, cities and local consumer groups. This enables Meesho to offer highly relevant products in line with the changing shopping behavior across India’s diverse consumer groups.

The company said PRISM powers personalized recommendations, product discovery, catalog understanding and vendor-development tools across the platform.