Flipkart Deploys 250+ AI Models and Custom LLMs — A Blueprint for Enterprise AI at Scale

Flipkart has quietly become one of the most aggressive AI deployers in e-commerce, with over 250 AI models now running across its platform and custom large language models (LLMs) generating 35-40% of the company's software code.

According to Chief Product and Technology Officer Balaji Thiagarajan, Flipkart is building what it calls an "agentic ecommerce platform" — a mix of frontier AI models and proprietary systems designed to handle everything from product discovery to seller support to back-office operations.

What sets Flipkart's approach apart is how it measures AI impact. Rather than tracking cost savings alone, the company monitors business metrics: click-through rates, conversion rates, average basket sizes, customer engagement, demand forecasting accuracy, and inventory turnover.

Key Takeaways:

• AI voice agents now handle approximately 90,000 personalized seller calls per month, with volumes expected to scale significantly.
• Flipkart integrates generative AI into product discovery, conversational shopping, catalogue creation, customer support, and internal developer productivity.
• The company is recruiting AI-focused leaders from Amazon, Tata Digital, Coupang, Razorpay, and Groupon.
• Rivals are moving fast: Amazon committed $48B to India cloud/AI infrastructure by 2030; Meesho reports over 70% AI-generated code.

The Flipkart model — business-metric-first AI tracking combined with custom LLM development — is a concrete framework for enterprise AI teams debating how to measure AI ROI beyond headcount reduction.

Read the full article on Apparel Resources