Reimagining the Bazaar: Can AI bring live commerce to India?

If an Indian from 2013 were to time-travel to 2023, among the many things they'd find unrecognisable is the average urban Indian's shopping cart. The emergence of large e-commerce platforms, increasingly efficient delivery systems, and a surge of local direct-to-consumer brands have greatly expanded the options available to Indian online shoppers. This upward trend shows no signs of slowing down. Despite these changes, most online purchases start with a search or are driven by click-through digital ads. These methods serve basic needs and offer a sense of value, and sometimes do serve spontaneous buys during social media browsing.

Yet, as a country that has rapidly moved on from the good ol' Fiat to Maruti/Honda to over a dozen global car brands and from Lakme & Lux to L'Oreal to now hundreds of scaled beauty brands, we believe Indian consumers will appreciate a more interactive and personalised online shopping experience.

That is where live commerce comes into play. A well-conducted livestream gives viewers a better understanding of the product while answering specific questions, providing additional information that can lead to more sales. The success of live commerce can be seen in China, where platforms like Taobao, Douyin, and Kuaishou made over $200B in sales from live streams last year, contributing to more than 10% of China's total e-commerce gross merchandise value (GMV)!

Recognising the potential of this alternate shopping path, top marketplaces like Myntra and Nykaa have begun to emphasise this feature in their apps. However, the proportion of live commerce in India's total e-commerce sector remains small, and startups have struggled to grow.

A key challenge for any upcoming live commerce platform is unit economics, which is only favourable for a country like India with a low average order value (AOV) if the platform achieves massive scale (thus ensuring a high no of concurrent users per stream). Picture this as an example: selling an INR 1000 item over livestream would require a significant investment in creator fees, logistics, video infrastructure, and marketing, totalling around INR 5-10k per stream. Just to break even, with the assumed 20% platform fees and a 1% conversion rate, a single stream would require 25-50k concurrent viewers. Add to it the complexity of doing numerous streams a day (that too of different types) to engage & retain the platform's users, and actual economics start looking much harder.

What changes with the advent of Generative AI?

Generative AI offers a direct solution to the unit economics problem. Advanced tools powered by robust AI models drastically cut the cost of content creation and open avenues for scalable personalisation and interactivity. Here's how we envision an AI-native live commerce platform:

1. AI-generated content

• Generative AI has democratised video editing, as evidenced by tools like Descript, Adobe Premiere, Pictory, etc. With further developments enabling real-time editing, such tools could exponentially boost commerce video content appeal through custom backgrounds, dubbing, lighting and frame correction, live script editing, and more. It could eliminate studio costs and offer numerous variations from a single recording customised to individual user preferences. Similarly, editing videos with virtual product placements can stop logistics costs associated with natural product shoots.
• In the long term, when AI-generated videos become indistinguishable from those shot on cameras, an unlimited variety of customizable video content will be possible. For example, companies like Runway & Synthesia have shown promising strides towards more life-like videos.

2. Enhanced product catalogues

• Many Gen-AI tools today can create bespoke product catalogues & descriptions for all products on the platform, making it more efficient to onboard the long tail of sellers.
• As a user spends more time on the platform, their preferences can be learned, allowing for personalised product descriptions. Thus, personalisation extends beyond product recommendations to customising how products are presented to each individual.

3. Real-time interaction for everybody

• Thanks to large language models (LLMs), it's now feasible to develop bots with human-like language skills to answer all live shopper queries, significantly boosting engagement.
• Real-time (relatively lightweight) AI models can highlight the most relevant questions for a human creator to answer, adding a gamification element to the browsing process.

The thoughtful application of Generative AI can resolve outstanding challenges in live commerce. With that backdrop, we're excited to connect with bold founders redefining India's shopping experience.