Why can an AI browser bring everything together?
When AI can accomplish most complex tasks with just a box and a browser interface, it also means that the only task left for the desktop is to support people in opening the AI browser.
When people entered the Internet era, driven by the needs for information production, sharing, and consumption, a relatively linear development path emerged: web pages - browsers - portals - search engines - social networks. Based on this path, as people embrace the AI era, different enterprises and developers use it as a reference and explore the possibilities of AI transformation at different nodes.
Among them, the browser was the first application accessible to the general public in the Internet era. Bill Gates wrote in his famous memo "The Internet Tidal Wave" released thirty years ago: "We believe that in the next 20 years, computers will have a fundamental impact on work, study, and play." Excellent software products are crucial for achieving these advancements. This is when Microsoft began to compete for control of the browser, the software that serves as the entry point.
In the AI wave, AI browsers have become one of the most fiercely competitive areas in AI applications. There are numerous participants in the competition for AI browsers, including star entrepreneurs, Internet veterans, domestic giants, and overseas up - and - comers. Fellou, Dia, and Opera are all searching for the best way to implement native AI browsers. Chrome and QQ Browser have also integrated AI assistant capabilities. OpenAI and Perplexity are also planning to launch their own browser products.
The entry point is the most direct allure that browsers offer. Every change in the mainstream browser represents a shift in control of the entry point. Microsoft and Netscape engaged in the first platform battle of the Internet era over browsers. Microsoft's victory further strengthened its dominance on the desktop through the combination of Windows and IE. When Google launched Chrome and dethroned the complacent IE browser, it also found the perfect partner for its search business.
When people realized that AI might bring about new disruptions, the browser remains one of the core components of the AI experience. However, AI browsers have transformed from containers for search to vessels for AI capabilities, and from windows for information acquisition to partners for proxy execution. Behind this change, AI is integrating search, decision - making, and execution, enabling people to perform more complex operations through natural language.
This also makes AI browsers an intersection point. AI applications in different directions, such as AI search, general agents, and copilots, will converge here. Doubao, an AI assistant, Quark, an AI search product, and ima, a knowledge - base product, all choose to present their capabilities in the form of browsers. Meanwhile, the virtual computers configured for AI by Kouzikongjian and Manus are more like collections of web - based tools built on browser capabilities.
Under this trend, AI browsers will become more like an operating system for the AI era. This was the goal that Google hoped Chrome could achieve - to bypass traditional desktop systems, use the browser as the core, and build a new software and hardware ecosystem with cloud capabilities.
When AI can accomplish most complex tasks with just a box and a browser interface, it also means that the only task left for the desktop is to support people in opening the AI browser.
With AI browsers, we can envision a simpler future interaction paradigm between humans and the world. This interaction paradigm will not only change software but also turn hardware into a provider of local computing power for AI and a connector to cloud computing power, creating a more flexible and lightweight software - hardware ecosystem. The former can fully utilize computer hardware configurations to support AI in performing complex tasks on the browser, while the latter can execute simple tasks on lightweight devices such as smart glasses.
AI - Driven Browser Transformation
Marc Andreessen, the founder of Netscape, once said, "The browser is the software that has had the greatest impact on people's lives in history." This statement still holds true today. Even in the era of widespread mobile Internet, where people are accustomed to searching for and browsing information on WeChat and Douyin, browsers still have hundreds of millions of users worldwide. They handle a large portion of complex information acquisition tasks and are the main way to open shared documents and multi - dimensional spreadsheets.
As AI becomes the driving force behind the new wave of technological innovation, browsers have become a key carrier of AI capabilities. Many released AI browser products have gained attention and popularity by offering smarter search experiences and more complex task - handling intelligence, creating the phenomenon of "invitation codes being hard to come by" time and time again. As of now, if you want to register and use AI native browsers like Fellou and Dia, you still need to obtain an invitation code first.
Driven by AI, although different AI browsers follow the design of an AI dialog box plus a browser interface, they have different focuses in actual use. Their different stances and starting points determine the development logic and user experience of the products.
We can generally classify AI browsers into three categories: AI native browsers, AI - like browsers, and AI - enabled traditional browsers.
Among AI native browser products, Dia makes me feel more like I'm using a regular browser, with a simpler and smoother operation. After entering content in its search box, users can choose to search and open web pages or start a chat conversation. On an already opened page, users can also start an AI conversation at any time to summarize and ask questions about one or more pages. Its workflow is to first meet the information acquisition needs and then transition to tasks such as writing articles or code.
In contrast, Fellou gives me the impression of being more like a Manus in browser form, with more complex functions. Due to its emphasis on proxy - based in - depth actions and proxy browsers, when using it, I immediately feel like I'm assigning tasks to assistants. After assigning tasks, I just wait for the assistants to complete them. This usage experience is exactly the same as that of Manus and Kouzikongjian, without a strong sense of browsing.
Beyond AI native browser products, the desktop versions of Doubao, Quark, and ima also exist in the form of browsers, but they don't highlight the browser's role in their product positioning. For these products that focus on AI assistants, AI search, and AI knowledge bases, the browser is more like an additional feature for users, allowing them to open task pages. Their product upgrades will still revolve around the optimization of AI for assistants, search, and knowledge bases.
In addition, traditional browsers are seizing the opportunities brought by AI to achieve a new lease on life. QQ Browser has launched QBot, which supports multi - modal search, automatic recognition of page - browsing intentions, and the invocation of office - assistance capabilities. Chrome has integrated the Gemini AI assistant in its updates, and Microsoft has launched the AI assistant Copilot Vision for the Edge browser. This change is essentially adding an AI plugin to the browser, enabling it to have more AI capabilities in addition to its original experience.
Compared with AI native browsers and AI - like browsers, traditional browsers face more constraints when undergoing AI transformation and find it difficult to quickly switch an old product to a new track. Therefore, they often choose a gradual transformation approach. Liu Hanbo, the person - in - charge of the QBot product, said that QQ Browser will achieve incremental improvements through AI while retaining its original high - quality business content. AI has indeed brought some incremental benefits. During the gray - scale testing phase before the official release of QQ Browser's AI features, the activity of the covered users increased, and the overall performance of the product improved.
The Competition Logic of Browsers: Entry Point, Operator, and Intelligent Partner
Although the implementation ideas of the three types of products differ, the underlying competition logic is the same.
First of all, it is a common understanding that the competition around AI browsers is about seizing control of the next - generation entry point. Josh Miller, the CEO of The Browser Company, the developer of Dia, said in an open letter that cloud revenues are skyrocketing, emerging startups are using browsers as platforms, the cryptocurrency ecosystem relies on browser plugins, and WebAssembly is opening up new interactive experiences...
Everything is converging towards the browser.
This convergence trend has become more pronounced under the influence of AI. People are increasingly relying on the web to meet their needs for AI search, AI image generation, and AI video generation. At the same time, we can also see a significant increase in AI social products on the web in the first quarter of 2025. E - commerce platforms like Taobao are also starting to pay more attention to the PC - side experience. People's interactions with Agent products like Kouzikongjian and Manus are also more concentrated on web pages.
Meanwhile, AI is changing the demand - matching model within browsers. Although traditional web pages are still used to carry these capabilities, in the future, people may not need to see traditional pages. More needs will be expressed in the browser's address bar, and then AI will allocate and execute tasks, and finally feed back the results in the chat interface extended from the address bar. Under this development trend, the browser's address bar will evolve into a super - box, similar to Quark's search box, and become the core entry point for AI.
Secondly, what everyone is competing for is not only the entry point for information acquisition but also the starting point for automated task flows. The browser will play the role of an operator in this process.
Arvind Srinivas, the CEO of Perplexity, believes that the Perplexity product is designed to complete a series of workflows. For example, Perplexity's Agent can help users manage investment portfolios. After reading reports and news, it can tell users how to adjust their risk exposure. This Agent needs to understand the entire process of search and information browsing, not just single - term searches.
The browser is a suitable carrier for this process. Arvind Srinivas said that if you really want AI to shift from providing answers to taking actions and doing things for you - answers are essentially a combination of four or five searches, while actions are a complete browsing session completed through a single prompt - you need a browser and seamless integration of computing power on the client and server sides.
According to the introduction by Xie Yang, the founder of Fellou, the top four usage scenarios of the Fellou product are information query, data scraping, social - media posting, and automatic form - filling. 60% of the use cases involve data query and scraping. Someone used Fellou to send job - seeking information to their desired companies on LinkedIn and received interview invitations from 10 companies, eventually getting hired by one of them. This shows that users indeed expect AI browsers to have stronger execution capabilities.
It should be noted that AI browsers can obtain more user context information through cookies, such as what users have browsed, what they have done, and which accounts they have logged in to on websites. At the same time, AI browsers allow users to access other third - party services through hidden tabs while logged in, scrape page data, and perform reasoning and execution operations on behalf of users. This can solve the problem that AI applications lack underlying control and cannot directly call other applications and obtain relevant information.
Finally, the browser competition may ultimately be a competition for intelligent partners.
In the future, browsers will become part of AI assistants. Xie Yang believes that what Fellou aims to do is to give more people freedom, more time, and the ability to be themselves. Josh Miller told The Verge that the core function of Dia is a chat tool that helps users find information and complete tasks more clearly and easily.
These positions all point to an intelligent partner with better memory and execution capabilities.
The Continuation of the Chrome OS Story
Based on my personal experience, current AI browsers are not yet competent intelligent partners; they are more like children who haven't learned how to use tools and professional skills.
We can ask AI browsers to write emails, send resumes, and create web pages on our behalf, but often the results are just mediocre, and some complex tasks may even fail during execution. For example, when using Fellou to create a news briefing for the day, it's difficult for it to obtain sufficient and effective information from news websites.
AI browsers need to have an ecosystem of capabilities and tools. To some extent, the main way for AI browsers to obtain information is no longer through browsing web pages but through direct communication with different applications and service providers at the data or protocol level. The AI browser Neon developed by Opera doesn't obtain information by browsing pages but analyzes web - page content through the DOM tree and layout data to acquire effective information more quickly.
This ecosystem may be divided into two layers. The first layer is to create a new AI plugin system based on the capabilities of AI browsers. Dia is trying to build such a skills ecosystem, allowing users to write code using Dia's "skills" function, create different tasks and capabilities, and distribute them like apps in the App Store, becoming a "routing system" that matches needs with capabilities.
The second layer is to connect with third - party platforms. According to a report by The Information in November 2024, OpenAI is also developing an AI browser product that will be deeply integrated with ChatGPT and will access the capabilities of service providers in tourism, food, real estate, retail, etc. Considering that Amazon is trying to create a shopping agent, we believe that the access to third - party capabilities will most likely be achieved through interactions between agents.
An AI browser with an application ecosystem will be more like an operating system. Josh Miller clearly stated that Dia ultimately aims to build a browser - based system that will become a powerful command center in a future full of agents. Arvind Srinivas also emphasized that the browser is essentially a containerized operating system and the best way to build agents.
This is very much like an AI - upgraded version of the Chrome OS story. In 2011, Google launched Chrome OS, which carried its ambition to develop an operating system based on web applications and cloud services. Later, Google also launched Chromebooks running Chrome OS. Unlike netbooks that use low - configuration hardware to run the Windows system for light - duty office work locally, Chromebooks combine cloud resources with Google services to meet users' daily needs.
The earliest Chromebooks were criticized as "browsers with a shell." However, as Chrome OS continued to evolve, Chromebooks have been making inroads in the education markets in North America and Europe, becoming the main choice for laptop consumers in the price range below $350. By this reference, if AI browsers truly become people's primary computing space, there may also be hardware products that match