One quick thing: Skyroot sets July 12- August 4 launch window for India's first private orbital missionIn today’s newsletter:Flipkart's e-commerce LLM pushSparrow Capital closes Rs 475 crore fundToken pricing comes to WhatsApp BusinessP.S.
You can sign up for Tech3 hereFlipkart's e-commerce LLM pushThe next e-commerce arms race may not be fought with bigger warehouses or faster deliveries.
CPTO Balaji Thiagarajan told us the company has deployed more than 250 AI models and is building an "agentic e-commerce platform" powered by both frontier and homegrown models.
It will be specialised e-commerce LLMs trained on Flipkart's own data and workflows.
As Thiagarajan put it, "General-purpose models won't be enough... That's the secret sauce."
One quick thing: Skyroot sets July 12- August 4 launch window for India's first private orbital mission
In today’s newsletter:
Flipkart's e-commerce LLM push
Sparrow Capital closes Rs 475 crore fund
Token pricing comes to WhatsApp Business
P.S.: Tune into Tech3 Podcast, your daily dose of tech and startup insights. Monday to Friday! Check it out on Spotify or Apple Podcasts. And don’t forget to sign up to The AI Edge, our weekly newsletter on all things artificial intelligence (AI).
Was this newsletter forwarded to you? You can sign up for Tech3 here
Flipkart's e-commerce LLM push
The next e-commerce arms race may not be fought with bigger warehouses or faster deliveries. It may be won one model, one agent and one line of AI-generated code at a time.
Beyond the bot: Flipkart isn't just plugging AI into its business. It's rebuilding the business around AI.
CPTO Balaji Thiagarajan told us the company has deployed more than 250 AI models and is building an "agentic e-commerce platform" powered by both frontier and homegrown models.
The real differentiator, he argues, won't be general-purpose AI. It will be specialised e-commerce LLMs trained on Flipkart's own data and workflows.
As Thiagarajan put it, "General-purpose models won't be enough... That's the secret sauce."
Code in cart: The AI shopping list extends well beyond coding assistants.
Around 35-40% of Flipkart's software code is already AI-generated, while the company is building in-house AI tools for engineers while also using Anthropic’s Claude Code and OpenAI’s Codex.
AI now powers everything from product discovery and personalised feeds to catalogue creation, customer support and Seller Lens. Voice agents alone make 90,000 personalised calls to sellers every month.
Scaling the stack: The biggest AI investment may not be in models alone, but in the people building them.
Flipkart has spent the past few months hiring senior AI leaders from Amazon, Coupang, Tata Digital, Razorpay and Groupon to deepen its technology bench.
Despite the industry's obsession with AI returns, Thiagarajan says Flipkart remains firmly in investment mode, focusing first on governance, safety and reliability before immediate ROI.
With Amazon pouring billions into AI infrastructure, Meesho generating over 70% of its code with AI, and Flipkart building its own specialised models, India's e-commerce battle is increasingly becoming an AI contest as much as a retail one.
