Browsers with AI: these are the best you can install

When generative AI is integrated into your browser

I’m a big fan of browsers. Always several of them installed on my computer and my mobile. Right now, I am mainly using Mozilla Firefoxbut it may not take long for it to change. For some time now, I have seen how AI has begun to gain ground in web browsing. Some integrations are useful, although lame. There are also those who take risks by launching agentic browsers.

I am going to talk to you about all this in the next paragraphs. My idea is not to propose a mere list of browsers with integrated AI. Actually, I want to explain which ones I have been trying and why I recommend each one of them.. Furthermore, I will tell you about one who stands firm on this matter and is resisting the trend of AI, refusing to integrate it. Shall we begin?

Three levels of integration

Before we get to work, I want to briefly talk to you about the three levels of AI integration in web browsers. Obviously, this does not meet any standard or agreed criteria, but rather what I have been able to experience in recent months.

For me, there are these three types of integration:

Agentic. There aren’t many of these, but they are starting to hit the market. I am referring to browsers that allow web browsing and tasks to be delegated to AI, without the user having to do anything more than send a request. They act as AI agents capable of making decisions and executing them on their own in the browser.
Attendance. It is the most classic integration. The chatbot on duty is integrated into a sidebar so that the user can answer questions about what they are seeing on the Internet. Some of these services connect very discreetly with navigation to be able, for example, to summarize the website, assist in writing texts and more.
Shortcut. This is the lowest level of integration. Basically, the browser allows the user to make quick queries to the AI from some area of the interface, but little else.

And, as I already told you, there are some browsers that don’t want to touch the AI with a stick, but I’ll save that for last.

Perplexity Comet is the browser of the agentic era

I’m going to start with one of the most surprising browsers I’ve tried lately. I mean Perplexity Comet. This is a proposal very far from what you already know. Basically, it is one of the first agentic browsers that any user can download. The idea is for AI to take control of navigation to execute complex tasks.

For example, this allows you to ask the chatbot, which is integrated into a side panel, to search for hotels on Booking, on specific dates, in a specific location and send the cheapest options to a spreadsheet. I was able to do something similar in the Comet analysis, which I invite you to read carefully.

Currently, this browser is experimental. It is because it is very slow and because the AI could still hallucinate in the middle of a task. Personally, I wouldn’t delegate any important tasks to the Comet chatbot. However, it is possible that this is an approach to web browsing of the future, where it is possible to let the browser work in the background while we do other things.

Comet is currently not available for mobile devices.

Dia Browser is one of the newbies

With permission from Comet, Day Browser It is one of the youngest browsers on the list. At the moment, it is only available for macOS. One detail that I didn’t like when trying it is that requires you to create an account. However, it has some interesting details.

Without going any further, it has memory. That means that it learns from the queries and tasks that you entrust to the AI. Thanks to these “memories”, the browser is capable of acting as a completely personalized assistant. Privacy is another issue.

Beyond that, the tools are similar to the rest. I haven’t been able to get the AI to act on the browser or the websites I’ve opened, so Dia is still in the assistance level.

Microsoft Edge, the pioneer of browsers with co-pilots

Microsoft Edge was one of the pioneering browsers in integrating AI. At first, everything was managed from Bing Chat, but then everything was in the hands of Copilot. We also dedicate a complete guide to this solution. In fact, I encourage you to review the best Copilot tricks in Edge.

Currently, Copilot on Edge provides you with the following services:

Side panel. A chatbot to resolve quick questions. It has the latest OpenAI models, so you can take advantage of it to use GPT-5 without limits. This side panel is loosely connected to the navigation, allowing you to invoke tabs (you have to enter an at sign and write the name of the open website), summarize content and ask questions using the content as context.
voice mode. Copilot in Edge also features voice mode, which lets you chat with the AI. This is useful, for example, to ask questions as you browse the web.
Copilot Writing Tools. In text fields, Edge’s AI helps you type, extending texts or converting them into lists. They are typical writing tools, similar to those of Apple Intelligence.
Ask Copilot. This function appears in the context menu after selecting text. It is also displayed when opening PDF documents with the browser.

Something notable is that this integration, in part, is found in the Edge mobile apps.

Mozilla Firefox is integrating AI despite criticism

Mozilla Firefox, which is the browser I use right now, has several versions integrating AI. The curious thing is that the most loyal users do not strongly agree with this strategy. The browser has even had performance issues due to AI processes.

Be that as it may, the truth is that many of the AI-based tools that Firefox now has have been useful to me. Yes indeed, all work with external services which the browser will consult to complete the different tasks.

Essentially, Firefox has a side panel where you can select a chatbot, such as ChatGPT, Gemini or Le Chat de Mistral. From there, the browser sends requests to summarize a website, to resolve doubts or to improve a text. The integration, in this case, is much more spartan. Even if you execute a task from outside the side panel, the query will be done right there.

In addition to this, Firefox is integrating small language models to solve other issues, such as intelligently group tabs. However, these are executed locally.

At the moment, Firefox for mobile does not include any of these features.

Brave has Leo

In addition to the examples that I have mentioned, there are many other small browsers that are doing the same and integrating AI into their interfaces. An example is Brave, which has its own service, Leo. In reality, it uses models from other companies, but focuses on offering privacy.

Leo is designed as a Browser native AI assistantallowing you to interact directly with web content in the same way Edge does: by reading the content of the current page to summarize articles, analyze documents or generate new text.

The grace of Brave is not in its functions, but in privacy. The company ensures that chats and requests are not used to train AI models, and that in addition They do not store your conversations or your instructions. Beyond that, there’s not much to highlight, beyond the fact that Leo is available in the versions of desktop and mobile of the browser and that has a clear focus on privacy. It’s one of the lightest integrations I’ve tried so far, with the exception of Chrome, of course.

Aria is Opera’s AI

Personally, Opera is one of the browsers that attracts me the least. Now, I know there are many users who love its extra features. Among them is Aria, the AI that runs in the browser’s side panel. The best thing is that it is integrated into the different versions of the browser, including Opera GX, which is presented with a gaming manual.

Among its functions is the image generationcommands to control tabs and quick access to AI with a keyboard shortcut. Another advantage that Opera highlights is that no account is necessary, although this is common in most cases.

In this case, Aria is available on all devices, including the mobile versions of Opera.

The curious case of Google Chrome

The last integration that I propose is that of Google Chrome. Although Google is one of the leading companies in AI, its browser barely has generative AI. In fact, access to Gemini is very weak.

The only thing you can do natively (I mean, without installing Chrome extensions) is write @gemini in the address bar and press tab. So everything you type will be sent as a prompt to Gemini.

It seems that this could soon change. As announced by Google, Gemini will soon be accessible in an exclusive panel. Additionally, many other features will be added, some of them related to user security. It seems that the idea is put Chrome at the level of Edgeallowing open tabs to connect with Gemini AI and execute tasks. Of course, a level of integration as brutal as that of Comet is not expected.

By the way, this integration does not work in the Android and iOS version.

There is a browser in Europe who doesn’t even want to hear about AI

And it has a composer’s name. Yes, the team Vivaldiwhich is a good European alternative to Google Chrome, is very clear that it does not plan to integrate AI. At least, that’s how it was in early 2024. So if you’re worried about your privacy or your browser being cluttered with features you don’t need, here’s a browser that acts as a haven. In addition, you can use it as an email client, RSS reader and calendar manager.