Google Turns Gemini Into a Checkout Layer for India

AI search has spent the last two years helping people decide what to buy. Google now wants it to help complete the purchase too.

Google has begun testing a new shopping experience in India that allows some users to purchase select Flipkart products directly through Gemini and Google AI Mode.

The limited test currently covers products such as smartphones, electronics and accessories. When eligible users see a supported Flipkart listing, a Buy button can take them into a Flipkart-branded checkout experience without leaving the AI interface. Google reportedly plans a wider rollout in October, ahead of India's festive shopping season.

That makes the experiment more important than another shopping feature.

Google is testing whether an AI interface can become part of the transaction itself.

Search is moving from recommendations to actions

Traditional search commerce has a familiar structure.

A user searches for a product.

Google shows links.

The shopper clicks through to a retailer.

The retailer owns the transaction.

Generative AI changes that flow.

A shopper can ask:

“Find me a good phone under ₹30,000 with strong battery life.”

The AI can compare products, explain trade-offs and recommend an option.

If checkout also happens inside the same experience, the journey gets dramatically shorter.

The AI moves from being an information layer to becoming a commerce layer.

That is a much more valuable position.

Google has been preparing for agentic commerce

The Flipkart test fits into Google's broader work around the Universal Commerce Protocol, an open standard designed to let AI agents communicate with retailers and support steps across a shopping journey, including transactions.

That is important because the future of AI shopping probably cannot depend on Google building a separate integration for every retailer.

Agents need a common way to understand:

products,

availability,

pricing,

checkout,

and order information.

Commerce standards could play the same role APIs played in earlier software platforms.

India is a strategic testing ground

India gives Google an unusually important environment for testing AI commerce.

It is one of the world's largest internet markets, with Flipkart and Amazon competing heavily for online shoppers, especially during festive-season sales. Google also has a financial relationship with Flipkart after investing roughly $350 million in the company in 2024.

That relationship gives Google a major retail partner with enough scale to test whether users are comfortable buying through AI.

If the experiment works, similar integrations could expand to other markets and merchants.

Retailers face a new discovery problem

For online retailers, AI shopping creates an opportunity and a threat.

The opportunity is obvious.

A retailer's products can appear inside highly personalized AI recommendations.

The threat is that shoppers may increasingly interact with the AI rather than the retailer's own website or app.

That means the assistant could control:

which products are compared,

how features are summarized,

which options appear first,

and where the user eventually checks out.

Search-engine optimization spent decades shaping online retail.

Agent optimization could become the next battleground.

AI shopping also changes advertising

Google's core business has historically benefited from commercial searches.

Someone searches for running shoes.

Retailers pay to appear.

But if the AI directly recommends one product and completes the purchase, the traditional list of sponsored links becomes less central.

Google will therefore need to figure out how advertising, recommendations and transactions fit together without making AI answers feel like paid product placement.

That balance will matter enormously.

Trust is more important when an assistant is deciding what to buy than when a search engine is merely displaying links.

Amazon now faces another interface battle

Amazon has spent decades trying to become the default starting point for product searches.

AI assistants threaten that position.

If users start a shopping journey in Gemini, ChatGPT or another agent, they may never open a retail marketplace until the transaction is nearly complete.

That shifts power upstream.

The company controlling the agent could control product discovery.

The marketplace may increasingly become fulfillment infrastructure behind the assistant.

What happens next?

The Flipkart test is still limited, and Google has not confirmed how broadly the experience will expand.

But the direction is clear.

AI companies do not want assistants that merely tell users what to do.

They want assistants that can complete the task.

Travel booking.

Restaurant reservations.

Payments.

Shopping.

Each one moves AI closer to becoming the interface between consumers and the rest of the internet.

Google's Flipkart experiment may look like a small checkout test today.

It could be an early preview of a much bigger shift:

The future of online shopping may begin with a conversation, not a search box.