I spent the morning from Gemini in Chrome, a fresh integration that places an AI assistant in the browser. Instead of going to the chatbot online application, you can click the fresh Gemini button in the upper right corner of Chrome to start the conversation-but the key difference is that the built-in browser assistant can “see” what is on the screen while moving on the web.
For me, the integration of Gemini from Chrome seems to be the beginning of Google’s mission to make my artificial “agent” intelligence because I found that I want him to do it more than in reality. For now, you can try the Early Access of Gemini in Chrome, if you are AI PRO or AI Ultra Subscribent, and exploit the Chrome or Canary Version of Chrome or Canary version.
I started by using Gemini to summarize some articles The VergeAnd also find even news related to games on the main page, where she indicated the fresh Game Boy Games Nintendo added to the Switch Online Switch, upcoming upcoming service Elden Ring Film adaptation and a massive Valve steam deck update.
But Gemini can “see” only what is on the screen, so I discovered that if you want to summarize some elements such as The VergeSection of comments in the comments, you must see them before chatbot answers. The twins will follow you when you switch bookmarks, but it can only get information from one at once.
If you do not want to write, Gemini in Chrome also allows you to go to its “live” function by choosing a button in the lower right corner of the dialogue. From there, you can simply ask a question, and Gemini will answer the conversation.
I found this particularly useful to exploit next to YouTube movies, in which I raised the film to the reconstruction of the bathroom and asked: “What tool does it use?” Gemini replied: “It looks like he is using a nail pistol to attach some wood.” In another film, Gemini correctly identified the capacitor on the motherboard, along with the tweezers and the sizzling air tool, youtuber used to remove it. He can summarize movies and talk about specific parts that you have not watched either, but I discovered that it is not always right if the video does not have marked chapters from which it can get information.
Probably my favorite exploit of integration is the recipes of Gemini Pull from YouTube movies, so I did not have to write recipes or look for a link in the description. He was also useful when I asked him to indicate waterproof bags on the Amazon search page.
The twins were not always consistent. When I asked Gemini, where is Mrbeast during The film of his discovery of ancient Mayan citiesIn this chichén itzá, he replied: “I do not have access to information in real time, so I cannot indicate the exact current location of Mrbeast.” When I asked about it again, the location mentioned in the film description: Mexico. Another time I asked Gemini to link to buy a specific Piępc pair shown in the film, but Gemini again told me that he has no access to real -time information, including offers of products or store stocks.
Sometimes I felt that Gemini’s answers were simply too long for a miniature window jumping in chrome. You can extend it, but it does not leave much space on the 13-inch display of my MacBook Air. In addition, one of the main AI sales points is that it is to facilitate you save time, giving quick and concise answers, which I do not always do unless I ask for it. Gemini’s subsequent questions, such as this, or if I would like to learn more about a given topic, have also become a bit repetitive.
Even in the case of these hiccups, I easily see Google to expand Chrome’s Gemini beyond ordinary questions and answers. Google wants AI to become “aggressive”, which means that it can perform tasks on your behalf, and the twins in Chrome seems to be ready to take this type of function. After asking Gemini to summarize the restaurant menu, for example, I even thought about placing an order for a pickup – an aggressive task that he simply can’t do. In the future, I could even see how it is useful when wearing tabs related to travel tests, and maybe even finding and saving movies on YouTube various recipes for my later playlist.
It seems that Google is approaching the realization of this with the “Agent mode” Project Mariner coming to the Gemini application, which will allow her to manage to 10 tasks at the same time and search on the web – or maybe one day she will bring these opportunities to Gemini in Chrome.