Remember when shopping online meant opening six tabs? You’d type a vague description into Google, get a wall of links, click three of them, compare prices in your head, lose track of which tab had the good deal, then give up and buy whatever was cheapest on the site you already had an account with. That ritual has been the default for about two decades.
Google is now testing something that skips most of it. In India, the company is trying out shopping directly through Gemini and AI Mode in Search, with checkout handled by Flipkart, the Walmart-owned retailer. The test is small right now, covering select products and select users, and a wider rollout is planned for later in October.
What is actually happening here
Normally, an AI assistant that helps you shop is doing something pretty modest. It answers questions, summarizes reviews, maybe suggests a few options, and then hands you a link. You’re still the one who opens the retailer’s site, logs in, picks a size, and taps buy. The assistant is a research helper, not a shopper.
What’s being tested in India moves the handoff point much later. You ask about a product inside Gemini or AI Mode, and instead of being dropped at a search results page, you land at Flipkart-branded checkout. The retailer is still the one selling to you and still the one you’re buying from. The difference is that the conversation and the transaction now live closer together.
For readers of this site who are still getting comfortable with what “AI agent” means, this is a useful example to hold onto. An agent isn’t just a chatbot with better manners. It’s software that takes steps on your behalf inside real systems. Checkout is one of those steps, and it’s a meaningful one, because it involves money, addresses, and payment details.
Why the standard matters more than the feature
The more interesting part isn’t the Flipkart test itself. It’s the plumbing underneath it. Google has described an open standard designed to let AI agents interact with retailers across the shopping journey, checkout included. The company said the technology would let shoppers buy eligible items through that flow.
An open standard is a boring phrase that does a lot of work. It means the connection between an assistant and a store isn’t a one-off custom deal that took a year of engineering and a signed contract. It means there’s a shared way for agents and retailers to talk, which in theory lets many assistants work with many stores without rebuilding the pipes every time.
If you want a sense of where AI agents are headed, watch standards, not demos. Demos show what one company can do with a partner. Standards hint at what thousands of companies might do with each other.
Why India, and why Flipkart
The verified details don’t spell out Google’s reasoning, so I won’t pretend to know it. What I can say is that testing a payment-adjacent feature in one market with one large retail partner is a sensible way to find out what breaks. Shopping is messy in ways that vary by country: payment methods, delivery expectations, return policies, product catalogs, tax rules. A small test tells you where the sharp edges are before millions of people find them for you.
The Walmart connection is also worth keeping in view. Flipkart is Walmart-owned, which means a US retail giant’s Indian arm is among the first to plug into Google’s assistant-driven checkout. Retail partnerships like that tend to shape what the standard looks like in practice.
What to watch as a regular person
You don’t need to follow the engineering to follow the story. A few practical questions are worth keeping in mind as this expands:
- Where does the buying decision actually sit? A flow that takes you to a branded checkout is different from one where an agent completes a purchase on its own.
- How clear is the price and the seller? Convenience gets uncomfortable fast when you can’t tell who you bought from.
- How easy is it to back out? Returns and cancellations are the real test of any shopping experience.
- Which stores show up? If assistants recommend products, catalog coverage starts to matter a lot.
A small test with a long shadow
It’s tempting to treat this as a minor product update. Limited users, limited products, one country, one retailer. Fair enough. But the direction is clear enough to name: assistants are moving from telling you about things to helping you get them, and checkout is the line that separates those two jobs.
Crossing that line is where AI agents start touching your wallet instead of your curiosity. That’s the part worth paying attention to as October brings a wider rollout, and it’s the part I’d keep an eye on well after the novelty wears off.
🕒 Published: