Meesho PRISM AI drives 75% of orders for 264 million users
Meesho disclosed that its PRISM AI system drives 75% of orders for 264 million annual transacting users by shifting from keyword-led search to real-time intent understanding. The system processes 6 trillion daily inferences using 100+ AI models and supports 10+ languages. The platform recorded 717 million placed orders in Q4 FY26.

*this image is generated using AI for illustrative purposes only.
Meesho, India's largest e-commerce platform by Annual Transacting Users and Placed Orders, revealed on June 04, 2026, that its proprietary AI intelligence system, PRISM, now drives 75% of orders by shifting commerce from keyword-led search to real-time intent understanding. The system supports 264 million annual transacting users and 717 million placed orders in Q4 FY26, processing over 6 trillion inferences daily to facilitate product discovery for users who browse rather than search with fixed intent.
PRISM Architecture and Scale
At the core of PRISM is a real-time intelligence and ranking architecture built to understand how Bharat shops online. The system interprets behavioural, transactional, and contextual signals across users and their browsing journeys. It is powered by a network of more than 100 AI ranking models and trained on over 400 trillion input signals.
| Metric | Value |
|---|---|
| Annual transacting users | 264 million |
| Placed orders in Q4 FY26 | 717 million |
| Daily inferences | 6 trillion |
| Peak inferences per second | 100 million |
| AI ranking models | 100+ |
The system supports 10+ multilingual experiences across Hindi, Bengali, Marathi, Tamil, Telugu, Kannada, Malayalam, Gujarati, Punjabi, and Odia. It is powered by BharatMLStack, Meesho's in-house ML infrastructure platform built to support high-throughput AI workloads at lower inference costs than conventional cloud infrastructure.
Intent-Driven Discovery
Through Trendpulse, its LLM-powered discovery engine, PRISM interprets emerging demand patterns across regions, cities, and local consumer clusters. This enables Meesho to surface highly relevant products aligned with evolving shopping behaviours of India's diverse consumer cohorts. The platform's ecosystem scale includes 263 million monthly active users, 17 billion product views per day, 1,592 million ratings, and 505 million reviews.
Debdoot Mukherjee, Chief Data Scientist, Head of AI and Demand Engineering, Meesho, stated that the next hundred million Indians coming online will discover rather than search. He emphasized that this shift demands infrastructure built from the ground up for a reality where users speak, browse, and expect technology to meet them where they are.
Historical Stock Returns for Meesho
| 1 Day | 5 Days | 1 Month | 6 Months | 1 Year | 5 Years |
|---|---|---|---|---|---|
| +0.37% | -11.72% | -23.54% | -2.43% | -2.43% | -2.43% |
How will Meesho monetize the shift from search to intent-based discovery to increase average order value?
Could the BharatMLStack infrastructure be licensed to other enterprises to create a new revenue stream?
What measures are being taken to ensure the multilingual AI models mitigate bias across diverse regional demographics?


































