Labs

Google's Flipkart 'Buy'
Button Test in
Gemini Splits AI
Search Into Two
Gates

A limited India pilot puts a Flipkart-branded checkout inside Gemini and AI Mode for smartphones and electronics — while Amazon listings on the same answer get only a link. Ranking and transactability are now separate problems.

Google began testing a “Buy” button on select Flipkart product listings inside the Gemini app and AI Mode in Google Search for shoppers in India, TechCrunch reported on September 26. Tapping the button opens a Flipkart-branded checkout flow without leaving the AI interface. On the same answer surface, listings from rivals including Amazon appear beside Flipkart’s, but with only a link out, no in-answer purchase path.

That asymmetry is the story. The catalog is narrow (smartphones, electronics, mobile accessories), the audience is limited to some users, and a broader rollout is planned for later in October, ahead of India’s festive shopping season. As FourWeekMBA put it, a pilot scheduled into a retailer’s busiest weeks reads as a revenue test rather than a research one.

The plumbing is opaque. Google introduced its Universal Commerce Protocol earlier this year as an open standard designed to let AI agents interact with retailers across the shopping journey, including checkout. But India hasn’t appeared on Google’s published UCP checkout rollout list, which so far names the US, with Canada and Australia next and the UK later. The Flipkart flow may run on something else entirely. TechCrunch itself notes it’s “not clear what technology powers the test.” A Google spokesperson offered only that the company is “always testing new features and experiences to help people discover and connect with businesses more easily.” Flipkart didn’t respond to TechCrunch’s request for comment.

Walmart owns Flipkart. Amazon is the incumbent Flipkart has fought for a decade. Handing one side a checkout button and the other side a hyperlink, on the same query, on the eve of festive quarter, isn’t a neutral experiment.

The structural read comes from FourWeekMBA: in a classic search result, ranking was the only gate; in an agentic interface, ranking and transactability come apart. Appearing in the AI answer is problem one. Being buyable inside the AI answer is problem two. Merchants that solve only the first watch the customer bounce out at the moment of intent.

Two adjacent moves this week point the same direction. Amazon has opened its seller backend to AI agents, collapsing catalog operations into agent-readable surfaces. Pinterest has turned visual search into a performance ad channel, collapsing discovery into transaction. The Flipkart test is the same compression from a third direction.

Historically, platforms open transactional real estate to a favored partner first and standards second. Amazon’s 2017 Alexa Skills marketplace worked that way; Meta Shops rolled out unevenly through 2020 and 2021. The winners weren’t the merchants with the best listings. They were the merchants who were plumbed in before the button shipped.

Sources