Gemini in Chrome can now do more than answer questions in India. On October 7, 2026, Google brought its auto browse agent to Chrome on desktop and Android for Google AI Pro and AI Ultra subscribers in India. The agent opens websites, fills forms and adds items to carts while the user watches.

That changes who is using your website. Some of your visitors will soon be software acting for a person. At Debugged Pro, we build websites and apps for clients in India and the USA, so we read Google's announcement, its help pages and The Verge's coverage. Here is what the launch means and what we would check on any client site this month.

Header image from Google's India blog post announcing Gemini in Chrome with auto browse. Image: Google.

What Google Launched With Gemini in Chrome in India

The news came in a post on Google's India blog by Charmaine Dsilva, Senior Director of Product Management for Chrome. It covered two separate changes.

Gemini in Chrome on Android for everyone in India. The assistant reached Chrome on desktop in India earlier this year. Now all Android users in India get it too.
Auto browse for paying users. The agent feature is live on desktop and Android for AI Pro and AI Ultra subscribers in India.

The basic assistant can summarise long articles, answer questions about the page you are on and connect with Google apps like Keep and Calendar. It can also edit images with Nano Banana, Google's image model.

Auto browse is the bigger deal. According to Google's announcement for India, it can book parking for an event, fill out a long form or order groceries for the week. Google says the feature is "secure-by-design" and that its models are trained to spot threats like prompt injection.

Who Can Use It Right Now

There are a few limits worth knowing before you plan around it:

Paid plans only. Auto browse needs a Google AI Pro or AI Ultra subscription.
English first. Auto browse currently works only on devices set to English. Gemini in Chrome on Android starts in US English and UK English.
Daily caps. Google's help page lists up to 20 multi-step task requests a day on AI Pro and up to 200 on AI Ultra.
Not everywhere. It does not work in Incognito mode or on iPhone and iPad. Work or school accounts need an admin to switch it on.

So this is not every Chrome user in India yet. It is the paying, English-language, often high-intent slice of users. For many businesses, that is the exact group that books, buys and fills out lead forms.

How Gemini in Chrome Auto Browse Uses a Website

It helps to understand the flow, because each step touches your site in a different way. Based on Google's help documentation, a task goes like this:

The user opens the Gemini panel in Chrome and types a task.
Gemini shows a plan. The user checks that it understood the request and the personal details involved.
The user clicks Start Task. Gemini then picks which sites to use and works in a tab marked with an auto browse icon.
When it needs the person, such as for a sign-in or a sensitive step, it sends a notice at the top of the browser.
The user can take over the task at any time, then give it back or stop it.

One line in the help page stands out for site owners. Gemini chooses which sites to use, and it may share the user's personal info with those sites. With permission, it can also sign in using saved passwords.

Chrome auto browse plan screen next to a Google Form for cricket tournament registration, with Start task and Cancel buttons — Gemini in Chrome shows its auto browse plan before it fills a Google Form. Image: Google, via The Verge.

The screenshot above shows that plan step. Gemini reads a team roster from another tab and plans to fill a cricket tournament registration form. The panel also says the user is responsible for Gemini's actions during tasks.

This Is Not New in the US

Auto browse first launched for AI Pro and Ultra subscribers in the US in January 2026. As The Verge reported on the India expansion, it handles jobs like shopping, filling out forms and researching hotel prices. Earlier Verge coverage noted that Google is competing with other agent browsers, including OpenAI's Atlas and Perplexity's Comet.

For agencies with US clients, that means the agent has been using American websites for months. India now joins that group, and India has a huge mobile, Android-first audience.

Why Gemini in Chrome Matters for Businesses in India and the USA

Most websites were built for a person with a mouse or a thumb. An agent reads the page differently. It looks at the page structure, the labels and the buttons, then decides what to click.

When a site is messy, people often push through. An agent may stop, pick the wrong option or hand the task back to the user. In every one of those cases, you risk losing the sale or the lead.

Here are the kinds of businesses we think should pay attention first:

E-commerce and quick commerce. Google's own India demo shows auto browse adding ingredients for gulab jamun to a cart on Flipkart Minutes.
Booking sites. Parking, travel, restaurants, clinics and events all depend on forms and date pickers.
Lead generation sites. B2B companies, coaching institutes and real estate firms live on enquiry forms.
SaaS products. Sign-up and onboarding flows are long forms by another name.

Phone screen showing Gemini auto browse working on a Flipkart Minutes grocery page with a Take over task button — Google's demo of auto browse in India adding gulab jamun ingredients to a Flipkart Minutes cart. Image: Google.

We wrote earlier about why AI agents get blocked by websites and the trade-offs of letting them in. That debate is still open. But with Gemini in Chrome, the agent runs inside the user's own signed-in browser. To your server, it looks much like your customer.

An Agent-Ready Website Checklist

The good news is that most fixes are plain good web practice. They help screen reader users, slow connections and AI agents at the same time.

1. Use Real Buttons, Links and Labels

Agents rely on the meaning in your HTML. A clickable div with no label tells them very little. A real button that says "Add to cart" tells them exactly what it does.

Use button and a elements for actions and links.
Give icon-only buttons a clear accessible name.
Label every form field. The W3C forms tutorial covers labels, grouping and error messages in plain terms.

2. Make Forms Short and Predictable

Long forms are one of Google's headline use cases. That means agents will be filling yours.

Use standard input types for email, phone and dates.
Add autocomplete attributes so saved details fill correctly.
Show clear error messages next to the field, not just a red border.
Avoid custom date pickers that only work with a mouse drag.

3. Put Prices, Stock and Policies in Text

If your price or delivery date only appears inside an image, an agent may miss it. Keep key facts in plain text near the product or service.

Also keep fees honest and visible. An agent comparing options will compare totals, and surprise charges at the last step look bad to both people and software.

4. Keep Checkout and Confirmation Steps Clear

Google says auto browse asks the user before some sensitive actions, like purchases. Your site can make that hand-off smooth.

Show a clear order summary before payment.
Use one obvious final button, such as "Place order".
Make the confirmation page state what happened, with an order or booking number.

5. Watch for Prompt Injection on Your Own Pages

Google says its models are trained to detect prompt injection. That is when hidden text on a page tries to give orders to an AI agent. If your site shows reviews, comments or listings from users, someone could plant that kind of text.

Moderate user content, strip hidden or invisible text where you can and review what appears on high-traffic pages. Treat it the same way you treat spam.

6. Test Your Key Journeys With an Agent

The fastest way to learn is to try it. Pick your top three journeys, such as buy a product, book a slot and submit a lead form. Then ask auto browse to do each one and note where it slows down or asks for help.

Repeat the test on desktop and Android. Mobile layouts often hide buttons or labels that desktop shows.

Gemini in Chrome vs Classic Browsing: A Quick Comparison

Area	Human visitor	Auto browse agent
Reads the page	Looks at layout and images	Reads structure, labels and text
Fills forms	Types and fixes mistakes as they go	Uses details the user shared and saved data
Handles unclear buttons	Guesses and tries again	May stop or hand the task back
Final purchase	Clicks pay when ready	Asks the user to confirm sensitive steps
Sign-in	Types a password	Can use saved passwords with permission

What We Are Telling Our Clients

We are not telling anyone to rebuild their site for AI. We are telling them that the work they skipped on accessibility and clean forms now has a second payoff.

For a typical client, we would start with a short audit of the top journeys. Then we fix labels, buttons and form errors first, because those changes are cheap and low risk. Bigger changes, like a new checkout flow, come after testing shows where the agent really fails.

If you already use Google's AI tools at work, our piece on Gemini Enterprise and Indian IT partners covers the business side of that shift. If you want help with an audit or a rebuild, our software development consulting team can review your site and apps.

Key Takeaways

Gemini in Chrome auto browse is live in India for AI Pro and AI Ultra subscribers on desktop and Android, as of October 7, 2026.
Gemini in Chrome on Android is now available to all Android users in India.
Auto browse can fill forms, compare products, add items to carts and book services, with daily caps of 20 or 200 tasks.
The agent picks sites, may share the user's details with them and asks before some sensitive steps.
Clean HTML, labelled forms, visible prices and clear checkout steps make a site easier for agents and people alike.
Test your top journeys with auto browse now, before your competitors do.

About the Author

Nitin Kumar is the Lead Project Coordinator at Debugged Pro. He works with the design and development teams on websites, apps and AI projects for clients in India and the USA. He writes about practical AI changes that affect how businesses build and run their digital products.