Google is testing a direct purchase feature for Flipkart products inside its Gemini and AI Mode interfaces in India, moving its AI services from product discovery into actual transactions. A limited set of users now sees a "Buy" button on select Flipkart listings that opens a Flipkart-branded checkout flow without leaving the AI environment, according to people familiar with the matter and an experience viewed by TechCrunch.
The early test covers smartphones, electronics, and mobile accessories for a small group of users. Others continue to see standard Flipkart product listings in Gemini and AI Mode with no direct purchase option. Google plans to broaden the rollout later in October, ahead of India's festive shopping season, one person said.
A shift from answers to transactions
The experiment signals Google's push to make its AI services transactional, not just informational. Rivals including OpenAI are adding commerce capabilities to their own AI products, racing to move beyond answering shopping queries and recommending products toward playing a direct role in online purchases.
Google has been building infrastructure for this shift. Earlier this year, it introduced the Universal Commerce Protocol (UCP), an open standard designed to let AI agents interact with retailers across the full shopping journey, including checkout. The company said at the time that UCP would allow shoppers to buy eligible products through Gemini and AI Mode using a Google-hosted checkout experience. Google has since expanded UCP with features that let shoppers transfer items to a retailer's site to complete a purchase.
The Flipkart test differs from that earlier demonstration. Tapping the Buy button brings up a Flipkart-branded checkout flow rather than a Google-hosted one. Which technology powers the test remains unclear.
Google's stake in Flipkart
Google and Flipkart have a financial relationship alongside their technology partnership. Google invested roughly $350 million in the Walmart-owned e-commerce company in 2024 as part of a funding round, taking a minority stake. Earlier this month, Google named Flipkart among the merchants partnering to bring what it calls "agentic" shopping experiences to consumers in India, without disclosing details of the test or its timeline.
A Google spokesperson said the company is "always testing new features and experiences to help people discover and connect with businesses more easily," adding that the company regularly runs experiments and has no further details to share.
India's competitive e-commerce battleground
India is the world's second-largest internet market, with more than a billion internet subscribers. Flipkart and Amazon compete fiercely for online shoppers, a rivalry that intensifies during the country's festive season when e-commerce companies launch their biggest sales and promotions of the year.
For now, the Buy option does not appear across all retailers surfaced by Google's AI services. In the experience viewed by TechCrunch, listings from rivals including Amazon appeared alongside Flipkart products but offered no option to purchase directly through the AI interface. Flipkart did not respond to a request for comment.
Why this matters for executives and strategy
Google's test represents a concrete step toward AI interfaces becoming transaction endpoints, not just discovery tools. For product development and sales leaders, the shift raises strategic questions about channel control: if consumers complete purchases inside an AI layer rather than on a retailer's own site, the dynamics of customer data ownership, margin structures, and brand visibility change. Executives watching this space should track whether Google's UCP standard gains adoption beyond Flipkart, as that would signal a broader reworking of how AI agents interact with commerce platforms. Professionals evaluating AI's impact on their own customer journeys can explore structured learning paths through AI for Executives Courses or examine how automation fits into transaction workflows with AI Agent Courses.
Your membership also unlocks:
