Opera’s Operator will save you the clicks and browse the web for you

Date:

Share:


Opera web browser is entering the agentic era of AI tools. At MWC 2025, the company previewed a new feature called Browser Operator that can turn sentences into actionable commands for web surfing, instead of having users click through different pages and buttons.

All you need to do is write a query in natural language and the AI tool will handle the rest. For example, you can tell the Operator to book three concert tickets on a booking site. The AI browsing agent will select the right boxes, book the tickets, and land you on the final step where you just need to make the payment.

As the browsing agent kicks into action, users will be able to see the step-by-step breakdown of clicks and searches it performs, somewhat like an AI reasoning model. It can also handle multi-step queries, and remembers the context from previous conversations.

How does Operator work?

Nadeem Sarwar / Tech Reader

It’s not a fully autonomous browsing experience. Opera says it will stop at the steps where users might have to intervene, such as filling sensitive information like phone number or car details. At the heart of it all is the AI Composer Engine, which breaks down user queries into browsing instructions.

The best part about the Browser Operator is that it runs privately instead of a cloud environment, which means your data remains safe. “Opera’s Browser Operator runs natively inside your browser, on your device,” says the company. That means information such as cookies or browsing history never leaves your device.

Meet Opera’s Browser Operator – An AI that can do tasks for you on the web

Moreover, instead of looking at a web page as a picture or screenshot, it focuses on its textual representation. That means the AI engine doesn’t have to scroll through the whole webpage, speeding up the whole agentic process as a result.

Since it can interact with UI elements that are otherwise invisible to users, pop-up boxes and verification prompts don’t pose a hurdle. In case any intermediate steps need action, the Operator will ask for users to briefly take over.

Opera’s Browser Operator In action.
Nadeem Sarwar / Tech Reader

Opera is calling it a “human-in-the-loop” approach. For such situations, users can either do the needful on the webpage, or just type the necessary details in the Operator chatbox. At no stage, does the Operator save any information, as that is strictly for the web page to process and handle.

Opera’s Browser Operator is currently in the preview phase and will be released as part of a feature drop in the “near future.” This won’t be the first agentic AI product of its kind, however. OpenAI has also developed its own version of an autonomous tool called Operator, while Gemini Deep Research is another example of a web-based agentic AI product. 








Source link

━ more like this

HUAWEI Watch GT Runner 2 is the “it” Smartwatch for Marathoners

HUAWEI has returned this year with one of its greatest performance-centric smartwatch series; HUAWEI Watch GT Runner 2. Seamlessly running with both Android...

A $1,000 Xbox might actually make sense, if Project Helix gets it right

A $1,000 Xbox would have sounded downright absurd not long ago, but Microsoft’s next-gen system, reportedly called Project Helix, may be heading into...

AI chatbots that are fit only for adults are still appearing in kids toys

A new report from the U.S. Public Interest Research Group (PIRG) Education Fund has raised concerns about the growing use of artificial intelligence...

Anthropic research says AI can mass expose of anonymous internet accounts

New research involving scientists from Anthropic and ETH Zurich suggests that modern artificial intelligence systems could identify the real-world identities behind supposedly anonymous...

Phone-based system promises better avatar movement without expensive VR gear

Researchers have developed a new system that uses smartphones instead of specialized motion-tracking hardware to create more realistic avatar movement in virtual reality...
spot_img