Meesho AI in FY26: Vaani lifts conversions by 22%

Why Meesho is pushing deeper into AI-led commerce

Meesho is expanding its use of artificial intelligence across the marketplace, positioning AI as central to how consumers discover products and complete purchases. The company said its full-stack focus on accessibility, trust and transactability has helped lift conversion rates among new consumers, supporting faster growth in annual transacting users (ATUs). Founder and CEO Vidit Aatrey described the company as increasingly “AI-first”, both in product building and user experience.

Aatrey said more than 70% of Meesho’s code is now AI-generated, while platform experiments have doubled year-on-year. The company also said more than 75% of orders now originate from AI-driven personalised feeds rather than search. Meesho framed this as a structural shift from intent-led search to AI-led discovery, where the platform surfaces products based on signals and predicted intent.

Q4 FY26 operating scale: NMV and order growth

Meesho reported Net Merchandise Value (NMV) of Rs 11,371 crore in Q4 FY26, up 43% year-on-year. Orders in the quarter rose 43% year-on-year to 717 million, which the company attributed to continued onboarding of new users and deeper engagement from existing cohorts. Meesho also stated that it is India’s largest e-commerce platform by ATUs and orders placed.

The company said its net loss narrowed by 88% in Q4, without disclosing the absolute figure in the provided details. Management linked part of the improvement to AI-led efficiencies that support user experience and unit economics.

PRISM and Trendpulse: how discovery is being redesigned

On the recommendation side, Meesho highlighted PRISM, its in-house recommendation engine. The company said PRISM improved intent-aware discovery and feed relevance. This push matters because Meesho is increasingly relying on personalised feeds as the starting point for shopping journeys.

Meesho also described Trendpulse, an LLM-powered system it uses to proactively identify regional and cultural demand trends. The intent is to capture fast-changing preferences across geographies and languages, and then reflect those patterns in what users see.

Vaani: Meesho’s GenAI shopping agent launched in Q4 FY26

Meesho said it expanded its GenAI stack with Vaani, an AI shopping agent launched in Q4 FY26. The company reported that Vaani crossed 1.5 million users within its first month. Meesho also said adopters of Vaani saw a 22% conversion lift.

Vaani is positioned as an end-to-end purchase journey tool, supporting product discovery, comparison, payment selection and delivery confirmation within a conversational flow. In the material provided, Meesho contrasted this approach with voice agents that primarily assist discovery and respond to prompts.

Voice-led commerce and the “next 500 million” users

Meesho is explicitly targeting users in tier II and smaller markets who may find typing, filters and structured search less intuitive. The company framed Vaani as a way to replicate the conversational nature of offline shopping, allowing users to describe intent in everyday language and ask follow-up questions.

Meesho said the assistant is currently available in English and Hindi, with plans to expand to more Indian languages over time. In early usage, the company said voice-based search usage was previously about 10% of users, while nearly 40% of users with access to the assistant are now using voice to find products and navigate the shopping journey.

Seller-side automation: voice agents at marketplace scale

Meesho also described GenAI deployment on the seller side. The company said GenAI-powered voice agents were scaled to handle around 300,000 calls per day. These calls are meant to assist sellers through promotional and sale cycles, improve participation, and unlock new growth opportunities across the marketplace.

This focus on seller enablement sits alongside Meesho’s consumer-facing push, indicating that AI is being used to reduce friction across multiple operational layers.

GeoIndia LLM and TrustMesh: localisation and risk controls

Beyond recommendations and voice shopping, Meesho highlighted India-specific AI infrastructure. One example is GeoIndia LLM, a location intelligence model that the company said outperforms global geocoding systems for Indian addresses. Another is TrustMesh, a risk engine that blocked 9 million high-risk transactions in FY26.

Meesho positioned these systems as critical to operating in a market with complex address formats, language diversity, and higher trust barriers for digital shopping. The company’s framing suggests that conversion improvements are being pursued alongside controls that reduce failed or risky transactions.

Key reported metrics at a glance

Market impact: what changes when discovery shifts from search to feeds

Meesho’s update points to a clear change in how users enter the shopping funnel. If more than 75% of orders are coming through personalised feeds, the platform’s ability to predict intent and show relevant products becomes a core growth lever. This also increases the importance of recommendation quality, catalog relevance, and trust controls because the user is not necessarily searching with a clearly defined product query.

The company’s Q4 FY26 numbers provide the operating context for this strategy: NMV rose to Rs 11,371 crore and orders reached 717 million, both up 43% YoY. Meesho also linked AI-led improvements to stronger unit economics and said its net loss narrowed 88% in Q4.

Analysis: why Meesho’s “India-specific” AI stack matters

Meesho’s AI narrative is not limited to a single feature release. It spans consumer discovery (PRISM), trend detection (Trendpulse), assisted shopping (Vaani), seller support (voice agents), localisation (GeoIndia LLM), and transaction risk (TrustMesh). The company’s emphasis on vernacular and voice-first flows suggests it is optimising for users who may be new to online buying and need guidance beyond product search.

The metrics Meesho disclosed, especially Vaani’s 1.5 million users in a month and a 22% conversion lift for users who interacted with it, indicate that conversational interfaces can translate into measurable commercial outcomes when tightly integrated into the purchase journey.

Conclusion

Meesho’s FY26 updates show an e-commerce model increasingly shaped by AI-led discovery, conversational shopping, and India-specific infrastructure for trust and localisation. With Q4 FY26 NMV at Rs 11,371 crore and 717 million orders, the company is tying AI adoption to both growth and operating efficiency. Next signals to track will be Vaani’s expansion into more Indian languages and how Meesho scales its AI systems across a larger share of its user base.