For whom are AI browsers intended?

This week, OpenAI introduced ChatGPT Atlas, an AI web browser, prompting the question: Should we finally abandon Safari?

That announcement influenced our discussion on the Equity podcast’s latest episode with Max Zeff, Sean O’Kane, and myself, as we explored the browser market and some of its more obscure players. However, it seems none of us are planning a major shift anytime soon.

Sean mentioned that numerous companies have unsuccessfully tried to dethrone the dominant browsers, mainly because they can’t monetize the browser itself. This might not be as significant for OpenAI, given its increasingly substantial funding.

Max, who tested Atlas and other browsers boasting AI agents, reported only “slight efficiency gains.” He noted instances where the agent simply “clicks around on a website,” questioning whether this is something users genuinely desire, in addition to highlighting considerable security concerns.

Below is an edited and condensed preview of our discussion for clarity.

Anthony: I’m still using Safari, but I’ve been trying alternative search engines, distinct from Google, due to my frustration with the overabundance of genAI content at the top of my search results.

I also wonder, if these AI browsers gain traction, what will it mean for the concept of the open web? Although accessing web pages remains possible, it’s reasonable to anticipate websites becoming less relevant as AI interfaces and chatbots increasingly control our browsing experiences.

Max: I believe the agentic web has been a widely discussed concept, and I think it’s a truly interesting question. People have been brainstorming solutions to advance toward this perceived future. 

In some ways, it echoes previous tech trends, where the question arises: “What is the actual user experience? What tangible benefit does one of these tools offer to the user?”

And currently, it’s not that compelling. After experimenting with ChatGPT Atlas and Comet, the most charitable assessment is that they offer minimal efficiency gains. They might make you marginally more productive.

However, much of the time, you’re just observing it methodically clicking around on a website, executing tasks that I wouldn’t typically do. For example, having it find a recipe and add all ingredients to Instacart. I’ve never actually done that. It seems like tech enthusiasts always mention that example in their videos, but I wonder if it’s really that common.

There is a significant disconnect, as the tech industry emphasizes, “We’re developing these tools for the agentic web,” but what practical reason would a typical person have to use them? I’m not sure.

Sean: I haven’t used any of these [AI browsers], largely because I’m still quite traditional when it comes to browsing and search — a lot of my work requires document searches, which naturally involve navigating through specific sections of familiar web pages and conducting Boolean searches on Google. I might try them if Google actually removes Boolean search, which seems likely at some point, but we’re not there yet. 

What intrigues me about these AI browsers is that we’ve seen previous attempts by companies to compete in the browser arena, which consistently fail due to the inherent difficulty in generating revenue from a browser as a standalone product. Some have tried charging upfront, which provides temporary relief, but it’s ultimately unsustainable when competing with Safari, Chrome, or Firefox. 

What’s interesting is that you finally have companies with virtually unlimited resources, allowing them to persevere indefinitely, as they’re not immediately focused on profit. While they’ll likely monetize them eventually, OpenAI doesn’t need to see returns from this in the short term; they can simply release it and let it evolve.