Chrome auto-browse: How to use Gemini's new AI agent feature

Chrome auto-browse: How to use Gemini's new AI agent feature in 2026

Google just rolled out what might be the most significant Chrome update in years. On January 28, 2026, the company launched auto-browse, an AI agent powered by Gemini 3 that can navigate websites, fill forms, compare prices, and complete multi-step tasks without you lifting a finger.

This isn't your typical autofill. While Chrome has handled addresses and credit card numbers for years, auto-browse actually takes control of your browser to accomplish real tasks. Need to research flight prices across multiple date ranges? Done. Want to find that jacket you bought last year and reorder it? Auto-browse handles the legwork.

The timing isn't coincidental. Google is responding to a wave of AI-native browsers like OpenAI's Atlas (launched October 2025), Perplexity's Comet, and The Browser Company's Dia. Rather than asking users to switch browsers entirely, Google is retrofitting AI capabilities into the world's most-used browser.

For content creators tracking emerging AI features like this, tools like eesel AI blog writer can help turn breaking tech news into publish-ready blog posts quickly. It's particularly useful when you need to cover fast-moving topics while they're still relevant.

Key features in the Chrome auto-browse Gemini update

The January 2026 update brings more than just auto-browse. Here's what's new across the Chrome and Gemini ecosystem.

Auto-browse (the agentic feature)

Auto-browse is the headline feature, and for good reason. It performs multi-step tasks that would normally require you to click through multiple websites, compare options, and fill out forms manually. According to Google's official announcement, the feature uses Gemini 3's multimodal capabilities to understand text, images, and even video content on the pages it navigates.

Every action the AI takes is visible in real-time through the side panel. You can interrupt at any moment if something goes off track. The feature also integrates with Google Password Manager, so it can handle logins (with your explicit permission) when navigating between sites. The side panel displays each action as a numbered step in a real-time action log, showing which pages the AI visits, what it searches for, and what decisions it makes at each stage.

Gemini side panel

The old floating Gemini window is gone. In its place is a persistent sidebar that keeps the AI assistant visible while you work on your primary browsing tasks. The side panel shows a real-time action log when auto-browse is running and offers multi-tab awareness, grouping related tabs from the same session. This is available to all Gemini in Chrome users, not just paid subscribers.

Connected Apps integration

Chrome's Gemini can now pull context from across your Google ecosystem. This includes Gmail for reading emails and drafting responses, Google Calendar for scheduling context, Google Drive for document reference, Google Maps for location information, Flights and Shopping for travel and purchase context, YouTube for watch history, and Google Photos for image context.

This interconnection is what makes auto-browse more useful than standalone AI tools. When you ask it to book a flight for your upcoming conference, it can pull the conference dates from your calendar and your arrival preferences from past email conversations.

Nano Banana image editing

Google also introduced Nano Banana, a lightweight image generation and editing model that runs directly in the browser. You can transform images without downloading and re-uploading, useful for quick tasks like home redesign inspiration or creating simple infographics. Outputs are watermarked, and this feature is available to all Gemini users (free tier included).

Personal Intelligence (coming soon)

Later this year, Chrome will add Personal Intelligence, a feature that learns from your past Gemini conversations and pulls context from your entire Google ecosystem. This is opt-in, and users will control which apps are connected. Google says it will arrive "in the coming months."

Chrome auto-browse pricing and subscription tiers

Auto-browse isn't free. You'll need a Google AI Pro or AI Ultra subscription to access it, and it's currently US-only.

Source: Google One AI Premium pricing

A few things to note about these tiers. The AI Plus plan launched globally on January 27, 2026, at a promotional $3.99/month for the first two months. Despite being a paid tier, it does not include auto-browse access. AI Pro offers a free first month, which is probably the most sensible way to test auto-browse before committing. AI Ultra is clearly aimed at power users or developers, with 10x the daily task limit and $100 in monthly Google Cloud credits included.

The geographic restriction matters. Auto-browse is US-only at launch. Google hasn't announced an expansion timeline, but given their track record, expect other English-speaking markets first, followed by broader global rollout later in 2026. Platform requirements are Windows, macOS, or Chromebook Plus.

How to enable and use Chrome auto-browse

Getting started with auto-browse is straightforward if you meet the requirements.

Step 1: Check your subscription and location

First, verify that you have an active Google AI Pro or AI Ultra subscription. You can check this at one.google.com. You'll also need to be accessing Chrome from the United States. VPN usage hasn't been tested widely, but Google typically enforces geographic restrictions at the account level.

Make sure Chrome is updated to the latest version. The feature is rolling out gradually, so even if you meet all the requirements, it might take a few days before you see it.

Step 2: Access the Gemini side panel

Click the Gemini icon in your Chrome toolbar. This opens the new side panel on the right side of your browser window. The panel will show your conversation history and provide the input field for new commands.

If you don't see the Gemini icon, check that Gemini in Chrome is enabled in your settings. Navigate to Settings > Gemini and ensure the toggle is on.

Step 3: Give a natural language command

Type your request in plain English. Be specific about your parameters. For example, instead of "find me a hotel," try "find hotels in Austin for March 15-18 under $200 per night near the convention center." The more context you provide, the better auto-browse can narrow down options.

Good examples of commands include researching the cheapest flight dates between two cities for a specific month, finding and reordering a product you purchased previously, comparing prices for a specific item across multiple retailers, or filling out a form using information from an uploaded PDF.

Step 4: Monitor and confirm actions

Once you initiate a task, watch the action log in the side panel. You'll see each step the AI takes: which sites it visits, what it searches for, and what decisions it makes. At sensitive moments (anything involving money, social media posting, or account logins), auto-browse will pause and ask for your explicit confirmation.

This is important: the AI won't complete a purchase or post anything publicly without your approval. Google has built in what they call "confirmation checkpoints" at these critical junctures.

Tips for better results:

Be specific about constraints. Budget limits, date ranges, and preferences all help the AI narrow down options faster. Start with simpler tasks to understand how the AI behaves before trusting it with complex workflows. Keep an eye on the first few runs. Reliability varies depending on the websites involved.

What can the Chrome auto-browse feature actually do?

Google's announcement showcased several use cases that testers have successfully completed. Here's what works based on their documentation and early journalist testing.

Travel and vacation planning

Auto-browse can research hotel and flight costs across multiple date ranges to find budget-friendly travel windows. It pulls context from your Gmail (conference invitations, existing bookings) and Google Calendar to understand your schedule. You can ask it to find the cheapest week to fly to a destination, and it will actually check prices across different date combinations.

The feature can also draft emails to colleagues with your arrival times, pulling information from your confirmed bookings. This cross-app integration is where Chrome's approach has an advantage over standalone AI tools.

Shopping and e-commerce

This is the use case Google showcased most heavily. Auto-browse can re-order previously purchased items (finding that jacket you bought a year ago), identify products from inspiration photos and locate them online within your budget, apply discount codes automatically, and add items to cart across multiple retailers.

Google announced support for a Universal Commerce Protocol co-developed with Shopify, Etsy, Wayfair, and Target. This means smoother transactions on these platforms. For other retailers, the AI navigates their standard checkout flows.

You'll still need to confirm the final purchase. Auto-browse won't spend your money without explicit approval.

Administrative tasks

The more tedious your task, the more appealing auto-browse becomes. Google lists form filling, tax document collection from multiple sources, appointment scheduling, subscription management, and getting quotes from service providers as tested use cases.

One interesting capability: you can upload a PDF and ask auto-browse to fill out an online form using that document's information. This could be genuinely useful for repetitive paperwork where you're copying the same information across multiple forms.

Safety, privacy, and what to watch out for

Auto-browse sounds convenient, but handing your browser over to an AI raises legitimate questions. Here's what you should know.

Built-in safety features

Google has implemented several safeguards. Auto-browse pauses before any sensitive action: purchases, social media posts, and account logins all require your explicit confirmation. You can interrupt the AI at any moment by clicking the stop button in the side panel. All actions are visible and logged, so you can see exactly what the AI is doing.

Google also announced "new security defenses for agentic AI threats" in December 2025, though they haven't detailed exactly what these involve. The feature integrates with Google Password Manager, but only accesses credentials when you explicitly grant permission.

Privacy considerations

This is where things get more complicated. According to Ars Technica's technical analysis, all page content is sent to Google's cloud for processing. This is not local, on-device AI. The content is logged to your Google Account "temporarily" (Google hasn't defined how long "temporarily" means), and it's stored in Gemini Apps Activity if you have "Keep Activity" enabled.

The bigger concern: Google declined to clarify whether page contents will be used for future model training. When Ars Technica asked directly, Google didn't provide a clear answer. If you're browsing sensitive information (financial accounts, medical records, private documents), that content is being sent to Google's servers.

There's also the risk of prompt injection attacks. If you visit a malicious website, it could potentially manipulate the AI's behavior through hidden instructions on the page. Google says they've built defenses against this, but it's a known vulnerability with agentic AI systems.

Reliability concerns

Here's an honest assessment: browser-based AI agents are still unreliable. TechCrunch noted that "browser-based agents are finicky and often fail to complete tasks." Google's demos focused on shopping and travel, but "in real-world use cases, agents often don't get the intent or break during traversing different sites."

Ars Technica's take was similarly cautious: "Most of the computer-use agents we've seen so far have been sluggish and unreliable enough to require human supervision. That kind of defeats the purpose."

I'm skeptical about agentic AI tools designed to make your life more efficient. The bots are almost always overhyped, and I've found them to be consistently unreliable.

Wired

If you're considering auto-browse for critical tasks, start with low-stakes experiments. Watch how it handles simple requests before trusting it with anything important.

Chrome auto-browse vs. other AI browsers

Google isn't the only company betting on AI-powered browsing. Here's how Chrome's approach compares to the alternatives.

Source: TechCrunch competitive analysis

The strategic difference is significant. OpenAI, Perplexity, and others are building new browsers from scratch, asking users to switch from their current browser. Google is adding AI to the browser most people already use. Chrome dominates market share, so even a modest adoption rate for auto-browse translates to millions of users.

That said, purpose-built AI browsers might offer tighter integration and better performance for specific use cases. If you're comparing ChatGPT alternatives and AI tools more broadly, the browser you use matters less than whether the underlying AI model handles your specific tasks well.

Getting started with Chrome auto-browse

Auto-browse is promising but early. Here's how to think about whether it's right for you.

Who should try it now:

If you already have an AI Pro or Ultra subscription and you're in the US, there's no reason not to experiment. People with repetitive web tasks (regular shopping, form filling, research across multiple sites) have the most to gain. Early adopters comfortable with occasional failures will get value from learning the system now.

Who should wait:

If you're concerned about data privacy, particularly for sensitive browsing, hold off until Google clarifies their data handling policies. Anyone outside the US can't access auto-browse yet anyway. If you need 100% reliability for critical tasks, this isn't mature enough.

Next steps:

Check your Google subscription status at one.google.com. Update Chrome to the latest version. Try a low-stakes task first, something like research or price comparison rather than purchasing. Monitor the feature as it matures over 2026.

For teams creating content about emerging AI features, eesel AI blog writer can help scale coverage of breaking news topics. It took us from 700 to 750,000 daily impressions by publishing consistent, well-researched content. The tool is free to try, so you can see how it handles topics like this one before committing.

Q1: What is the Chrome auto-browse Gemini feature? A1: Chrome auto-browse is an AI agent powered by Gemini 3 that performs multi-step web tasks autonomously. Unlike traditional autofill, which only handles addresses and credit cards, auto-browse actually navigates websites, fills forms, compares prices, and completes purchases on your behalf. It launched in January 2026 and requires a Google AI Pro or Ultra subscription.

Q2: How much does Chrome auto-browse cost? A2: Chrome auto-browse requires a Google AI Pro subscription at $19.99/month (20 tasks per day) or AI Ultra at $249.99/month (200 tasks per day). The free tier and $7.99/month AI Plus tier do not include auto-browse access. AI Pro offers a free first month for new subscribers.

Q3: Is Chrome auto-browse available outside the US? A3: No, Chrome auto-browse is currently US-only at launch. Google hasn't announced a timeline for international expansion. The feature requires both a qualifying subscription and access from a US location.

Q4: What can Chrome auto-browse actually do? A4: Chrome auto-browse can research and compare travel options, re-order previously purchased items, find products within budget constraints, fill out online forms from PDF information, apply discount codes, collect documents from multiple sources, and schedule appointments. It integrates with Gmail, Calendar, and other Google services for context.

Q5: Is Chrome auto-browse safe to use? A5: Chrome auto-browse includes safety features like confirmation checkpoints before purchases, social media posts, and logins. However, all page content is sent to Google's cloud for processing, and Google hasn't clarified whether this data is used for model training. There's also potential vulnerability to prompt injection attacks on malicious websites.

Q6: How reliable is Chrome auto-browse? A6: Early testing suggests Chrome auto-browse, like other browser-based AI agents, can be unreliable. Tech journalists have noted that these agents "often fail to complete tasks" and may require human supervision. Google's demos focused on shopping and travel use cases, and real-world performance may vary across different websites and task types.

Q7: What's the difference between Chrome auto-browse and other AI browsers? A7: Chrome auto-browse adds AI capabilities to an existing browser most people already use, while competitors like OpenAI Atlas and Perplexity Comet are purpose-built AI browsers requiring users to switch. Chrome's advantage is its massive user base and deep Google ecosystem integration. Purpose-built AI browsers may offer tighter model integration for specific use cases.