Jess Weatherbed, reporting for The Verge:

The Browser Company CEO Josh Miller teased in October that it was launching a more AI-centric product, which a new video reveals is Dia, a web browser built to simplify everyday internet tasks using AI tools. It’s set to launch in early 2025.

According to the teaser, Dia has familiar AI-powered features like “write the next line,” — which fetches facts from the internet, as demonstrated by pulling in the original iPhone’s launch specs — “give me an idea,” and “summarize a tab.” It also understands the entire web browser window, allowing it to copy a list of Amazon links from open tabs and insert them into an email via written prompt directions.

“AI won’t exist as an app. Or a button,” a message on the Dia website reads. “We believe it’ll be an entirely new environment — built on top of a web browser.” It also directs visitors to a list of open job roles that The Browser Company is recruiting to fill.

The name “Dia” says most of what’s noteworthy here: The Browser Company’s next product isn’t a browser at all. It’s an agentic, large language model-powered experience that happens to load webpages on the side. Sure, it’s a Chromium shell, but the primary interaction isn’t meant to be clicking around on hypertext-rendered parts of the web — rather, The Browser Company envisions people asking the digital assistant to browse for them. It’s wacky, but The Browser Company has already been heading in this direction for months now, beginning with the mobile version of Arc, its flagship product. Now, it wants to ditch Arc, which served as a fundamental rethinking of how the web worked when it first launched a year ago.

The Browser Company’s whole pitch is that, for the most part, our lives depend on the web. That isn’t a fallacy — it’s true. Most people do their email, write their documents, read the news, and use social media all in the browser on their computer. While on mobile devices, the app mentality remains overwhelmingly popular and intuitive, the browser is the platform on the desktop. Readers of this website might disagree with that, but by and large, for most people, the web is computing. I don’t disagree with The Browser Company’s idea that the web needs to be thoroughly rethought, and I also think artificial intelligence should play a role in this rethinking.

ChatGPT, or perhaps LLM-powered robots entirely, shouldn’t be confined to a browser tab or even a Mac app — they should be intertwined with every other task one does on their computer. If this sounds like an operating system, that’s because The Browser Company thinks the web is basically its own OS, and it’s hard to argue with that conclusion. Most websites these days perfectly fit the definition of an “app,” so much so that some of the biggest desktop apps are just websites with fancy Electron wrappers. For a while, Arc had been building on this novel rethinking of the web, and while some have begrudged it, I mostly thought it was innovative. Arc’s Browse for Me feature, AI tab reorganization, and tab layouts on the desktop were novel, exciting, and beautiful. The Browser Company had something special — and that’s coming from someone who doesn’t typically use Chromium browsers.

Then, Miller, The Browser Company’s chief executive, completely pivoted. Arc would go into maintenance mode, and major security issues were found weeks later. It wasn’t good for the company, which once had a real thing going. I listened to his podcast to understand the team’s thought process and to get an idea of where Arc was headed, and I came to the conclusion that a much simpler version of Arc, perhaps juiced with AI, would come to market in a few months. The Browser Company had a problem: Arc was too innovative. So here’s what I envisioned: two products, one free and one paid, for different segments of the market. Arc would become paid and continue to revolutionize the web, whereas “Arc 2.0,” as Miller called it, would become the mass-market, easy-to-understand competitor to Chrome. It’s just what the browser market needed.

That vision was wrong.

Now, Arc and the stunningly clever ideas it brought are dead, replaced by a useless, flavorless ChatGPT wrapper. Take this striking example: Miller asked “Dia” to round up a list of Amazon links and send them in an email to his wife. The “intelligence” began its email with, “Hope you’re doing well.” Who speaks to their spouse like that? This isn’t a browser anymore — it’s AI slop. I understand the video and promotion The Browser Company published demonstrates a prototype, but writing emails isn’t the job of a browser. Search should be Dia’s main goal, and the ad didn’t even talk about it in any way that was enticing. Instead, it demonstrated AI doing things, something I never will trust a robot with. Booking reservations, creating calendar events, writing emails — sure, this is busy work, but it’s important busy work. Scrolling through Google’s 10 blue links is busy work that’s actually in need of abstraction.

This hard pivot from innovative ideas and designs to run-of-the-mill AI nonsense serves as a rude awakening that no start-up will ever succeed without ruining its product with AI in the process. Again, I don’t think it’s the AI’s fault — it’s just that there’s no vision other than venture capitalist money. A browser should stick to browsing the web well, and Dia isn’t a browser. There’s no place for a product like this.