E-commerce platform Meesho is betting on a vernacular voice AI assistant to make online shopping more accessible for new-to-e-commerce consumers, as it seeks to expand its annual transacting user base beyond 251 million.
The push comes at a time when India’s e-commerce growth is moderating. The market expanded at 10–12% in 2024, compared with over 20% historically, according to a Bain report published in March last year.
Meesho noted in its IPO documents in December that a majority of new online shoppers are expected to come from tier-II-plus cities, which are projected to account for 51–52% of India’s e-commerce market by FY30, up from about 44% in FY2025.
On Tuesday, the company launched “Vaani”, a generative AI-powered conversational voice assistant designed to replicate the interactive nature of offline shopping. Among peers, Amazon already offers an AI shopping assistant, Rufus, primarily in English, while Flipkart is reportedly working on a similar feature.
What did Sanjeev Kumar say?
“If you look at India’s internet population, we have roughly about 850 million internet users. So when we try to get the remaining 500 million online, connectivity is not really the problem, the bigger problem is the interface itself,” said Sanjeev Kumar, co-founder and CTO at Meesho.
The company is betting that conversational, vernacular, voice-led interfaces can help bring this next cohort of users online. Many of them — particularly in smaller towns — remain uncomfortable with app-based navigation centred on typing, filters and structured search.
In offline settings, shopping is inherently conversational, with customers describing their needs, asking follow-up questions and seeking reassurance before making a purchase. Vaani aims to replicate this behaviour online by allowing users to speak in Hindi and English (with more languages to be added soon), ask follow-up questions, and receive contextual recommendations, reviews and product details in real time.
AI stack expansion
Meesho has been steadily expanding its AI stack across the platform — from hyper-personalised feeds and seller advertising tools to customer support chatbots. In its IPO filing, the company earmarked ₹480 crore from the fresh issue for investments in AI and machine-learning teams.
The assistant is powered by an AI architecture that combines large language models fine-tuned on Meesho’s proprietary data, including regional language nuances. Within a month of launch, more than 1.5 million users had interacted with the assistant, the company said. Meesho also reported a 22% improvement in conversion rates among users engaging with the feature, along with lower returns and cancellations.