Scaling Development with Generative AI
SoftBank-backed Meesho has reached a significant milestone in engineering productivity, reporting that over 70% of its codebase is now AI-generated. CEO Vidit Aatrey highlighted that this integration spans the entire software development lifecycle, including coding, testing, reviews, and deployment. This shift has enabled the company to double its platform experiment rate in Q4 FY26, facilitating faster iteration cycles.
Transforming E-commerce with AI-Personalized Feeds
Meesho is shifting its user experience paradigm away from traditional search bars. Currently, 75% of orders originate from AI-led personalized feeds. This data-driven approach has yielded tangible business results:
- A 15% increase in conversion rates.
- A 27% reduction in the time required for new seller listings to gain market traction.
Pioneering Voice Commerce and Infrastructure
Beyond the feed, Meesho is investing heavily in accessible commerce and internal efficiency:
- Vaani: A new voice-based shopping assistant reached 1.5 million users in its first month, with adopters seeing 22% higher conversion rates, particularly among regional language speakers.
- Proprietary ML Stack: The company’s in-house infrastructure processes over 6.4 trillion inferences at peak, operating at 60–70% lower costs than standard cloud alternatives.
- Logistics: The proprietary ‘Geo-India’ model has improved geocoding accuracy by 20 percentage points, a critical advantage for delivery precision in smaller cities.
Strategic Takeaways for Founders
Meesho’s trajectory serves as a blueprint for e-commerce scaling: prioritize building in-house ML capabilities to reduce long-term infrastructure costs and lean into non-text search modalities like voice to capture the next wave of internet users in emerging markets.