Meesho’s new voice shopping assistant, Vaani, saw 1.5 million users activate it within its first month. This traction comes as companies like ShareChat and Rapido grapple with balancing advanced AI models against India’s low-end devices, patchy networks, and cost constraints. For product teams, it defines the path to truly mass-market AI adoption beyond the urban English-speaking user.
How We Got Here
While early AI adoption focused on high-end use cases, the conversation has recently shifted dramatically towards mass-market applicability for India. This was evident at a recent CTO summit in Bengaluru, where leaders from Meesho and ShareChat outlined strategies for reaching millions of users on budget devices.
The Numbers
- ShareChat/Moj CTO Nitin Jain advocates for infrastructure that allows companies to switch AI models and providers without rebuilding applications.
- Meesho reported 22% higher conversions among Vaani users compared to non-users in Tier III and IV cohorts.
- Flipkart, a competitor, is also integrating conversational AI, voice, and visual inputs for personalised product discovery, as outlined by CPTO Balaji Thiagarajan.
- Meesho is testing on-device AI to manage costs and evaluate its impact on memory, battery life, and user experience, especially on lower-end devices.
- ShareChat’s AI strategy is shaped by users relying on mid-range phones, limited bandwidth, and the need to support multiple languages with lower monetisation levels.
What Happens Next
🇮🇳 Why This Matters for India
For founders building SaaS or D2C products aimed at the next 500 million internet users in cities like Lucknow, Indore, and Jaipur, these AI strategies offer a playbook for unlocking significant market share through accessibility.
The Take
What’s being missed here is how deeply this pushes product teams towards a 'hardware-first' AI mindset, where phone specs dictate model design. The platforms that crack truly efficient on-device processing will own the next wave of Indian internet users, making cloud-heavy AI a niche.
Source:
Inc42 ↗