Meesho Launches Vaani — India’s First Vernacular AI Shopping Assistant for 250 Mn Bharat Users

E-commerce giant Meesho has launched Vaani, a multilingual AI voice shopping assistant designed for India’s next 200 million online shoppers — those who are more comfortable speaking in their native language than typing search queries in English. Built on generative AI, large language models (LLMs), and custom natural language processing trained on Indian vernacular patterns, Vaani marks the most ambitious AI product bet from any Indian e-commerce platform in 2026, arriving just months after Meesho’s IPO and its Q3 FY26 earnings that showed revenue up 32% YoY to Rs 3,517.6 Cr.

Vaani directly targets a structural barrier in Indian e-commerce: vernacular search and discovery. With voice search on Meesho growing 40% year-on-year and vernacular language usage up 162% on the platform, the product arrives with clear demand signals. It positions Meesho ahead of Amazon Rufus and Flipkart Immerse in the race to own AI-native commerce for Bharat — and Amazon and Flipkart’s predominantly English-first AI assistants will be watching closely.

What Vaani Does — Key Features

Vaani is not simply a voice search upgrade. It is a full-stack AI shopping companion built on Meesho’s multi-year investment in GenAI infrastructure, the Bharat ML Stack, and its ElevenLabs-powered voice synthesis layer:

• Vernacular voice shopping: Users can speak naturally in Hindi, English, or (in rollout) Bengali, Gujarati, Kannada, Marathi, Tamil, and Telugu. Vaani interprets colloquial queries, misspellings, and code-switched language inputs (e.g., Hinglish)
• Conversational product discovery: Unlike keyword search, Vaani engages in multi-turn dialogue — “show me a saree under Rs 500 for a puja” — mirroring how shoppers interact with an offline kirana store owner
• Low-device optimisation: Designed for basic Android smartphones with limited RAM and noisy environments (traffic, street noise, factory floors) — the same engineering constraints that shaped the customer support voice bot
• Post-purchase voice support: Vaani extends Meesho’s existing 60,000 daily-call GenAI support bot into the shopping journey — handling order status, returns, and refunds through the same interface
• Personalised feed integration: Vaani’s responses are grounded in Meesho’s Bharat ML Stack which processes 2 petabytes of data daily — recommendations are personalised to each user’s purchase history and browsing behaviour
• Agentic AI roadmap: Meesho AI Labs is piloting agentic shopping — where Vaani can complete a purchase end-to-end based on a single voice command, with payment and address confirmation through voice

Why Vaani Matters — The Strategic Case

Meesho’s entire growth thesis rests on democratising internet commerce for users who find English-first, text-heavy interfaces intimidating. CEO Vidit Aatrey has stated publicly that “there’s a digital divide between shopping online and offline” — and that AI, specifically voice AI, is the bridge.

The numbers validate the urgency: vernacular language usage on Meesho grew +162% in 2024, voice search on the platform grew +40%, and 1 in 3 buyers is under 25 — a generation that has grown up with WhatsApp voice notes, not typing. Vaani is engineered precisely for this cohort.

The strategic implication is profound. Meesho’s zero-commission model for sellers means its competitive edge cannot be built on price. It must be built on reach and discovery. Vaani extends Meesho’s discoverability to users who were previously excluded by language and interface friction — and that could translate to the company’s next 50 million transacting users.

How Vaani Compares to Competitors

Vaani’s deepest differentiation is vernacular depth and device-level inclusivity — Flipkart Immerse and Amazon Rufus are both compelling AI products, but neither is engineered from the ground up for users on Rs 8,000 smartphones in noisy Tier-3 environments. That is Meesho’s structural advantage, and Vaani is the fullest expression of it.

How Vaani Works — The Technology Stack

Notably, Meesho has not built its own foundational LLM — CTO Sanjeev Barnwal has been explicit that off-the-shelf models perform well for Hindi and English, and the company’s moat lies in its custom contextual layers, not model architecture. This keeps AI infrastructure costs manageable relative to Meesho’s unit economics.

What the Founders Say

“Our mission is how do we really democratise internet commerce for everyone, consumers and businesses. There’s a digital divide between shopping online and offline — and AI is the bridge.”

— Vidit Aatrey, Co-Founder & CEO, Meesho (Entrepreneur of the Year 2026 event)

Aatrey’s framing is deliberate. He is not pitching Vaani as a technology product — he is pitching it as a social equity instrument. This distinction matters for regulatory optics as Meesho navigates post-IPO scrutiny and for long-term brand positioning against Amazon and Flipkart, both of which are seen as urban-first despite their Tier-2 rhetoric.

“We have built India’s first GenAI voicebot at scale. When we continue to scale it, I suspect the cost will go down from 25% to maybe 15% of prior levels.”

— Sanjeev Barnwal, Co-Founder & CTO, Meesho

Barnwal’s cost focus reveals the business case clearly: every expansion of Vaani’s voice capabilities directly compresses Meesho’s cost-to-serve per user. At 250 million transacting users, a 1% improvement in cost efficiency is worth hundreds of crores annually.

Meesho Financial Snapshot — Context for Vaani’s Launch

The sharp widening of Q3 FY26 losses — from Rs 37 Cr to Rs 490 Cr — reflects Meesho’s deliberate investment in user acquisition and AI infrastructure. CEO Aatrey has framed this as investment-phase spending to onboard the next wave of internet shoppers. Vaani is part of that equation — reduce friction for new users, increase session depth, and improve repeat purchase rates without proportional CAC increase.

What’s Next

Meesho’s next AI milestone will be agentic commerce — where a user can say “order me a white kurta under Rs 400, size M” and Vaani completes the transaction end-to-end without any screen interaction. Meesho AI Labs is piloting this, and a public rollout is plausible by Q3 FY27 if the voice discovery metrics from Vaani’s initial launch prove out.

The commercialisation play is the less-discussed angle. Meesho Networks (set up April 2025) was created explicitly to commercialise AI technologies to third parties. Vaani’s underlying voice infrastructure — the ASR, TTS, NLP stack — could become a revenue-generating B2B product sold to other Indian consumer companies, transforming Meesho from a marketplace into an AI platform business. That is the real long-term prize.

The question investors will be asking: can Vaani move the needle on Meesho’s Q4 FY26 losses? The mechanism is clear — lower support costs (already down 75%), higher conversion from voice discovery, and lower customer acquisition cost per session. Whether the volume of adoption is large enough to show up in the P&L within two quarters is the variable to watch.