(Credit: Lily Yeh/PCMag; Microsoft/AI-asissted: Envato AI Labs)
I can’t imagine everyone in the world suddenly jumping off their current web browser to switch to an AI-powered one, especially since the zeitgeist is turning against AI. Microsoft offers a compromise, however, with the new Copilot Mode in Edge. In effect, Edge enables you to keep using a standard web browser while giving you the option to engage AI features at will.
You can find commentary on both sides of the AI browser fence. Some articles say, “I tried AI browser X, and switched right back to Chrome,” and others claim, “I tried AI browser Y, and now I can’t switch back!” This new Copilot mode in Edge saves you from the choice, at least if you’re among its nearly 300 million users—a large number that still pales in comparison with Chrome’s 3.7 billion users. It’s the browser that ships with Windows and is available for all major platforms.
Microsoft still presents the Copilot Mode for Edge as an experimental feature, but undaunted by that, I took the newfangled browser mode out for a spin. In short, I found some things to like and several shortcomings, but nothing showstopping in either direction. You should absolutely try it if you enjoy testing the latest tech, but just keep your expectations in check.
How to Get Copilot Mode in Edge
Copilot Mode is available only on the desktop version of Edge that runs on macOS or Windows (both 10 and 11). You also have to live in one of the 170 worldwide markets where Copilot is available. Copilot Mode is free, at least for now.
The first step is to update Edge to the latest version. In my case, I got a message saying that I “may need to” update to v.138.0.3351.109. Better safe than sorry. To update Edge, click the three-dot menu at the top right, choose the Help and Feedback option at the bottom, and then select About Microsoft Edge. You need to restart the browser to install updates.
Once you have the latest version running, you need only to go to the Copilot Mode in Edge site and slide the Turn On Copilot Mode switch to the right.
(Credit: Microsoft/PCMag)You can also just go into Edge’s Settings page, find the AI Innovations section, and turn on the Copilot Mode from there.
(Credit: Microsoft/PCMag)What Can You Do With Copilot Mode in Edge?
At the time of publishing, Copilot Mode in Edge offers just a few distinguishing features. In order of importance, these include a new AI-chatbot-focused new tab page, the Quick Assist Copilot button, voice-driven browsing, Copilot Actions, and multi-tab analysis. I take you through each of these capabilities below.
New Tab Page
When you open a new browser tab, you get an AI-centric page, rather than the standard one with a Bing search box, a daily photo, and news summaries. It’s nearly identical to what appears on the Copilot website itself, with a few minor differences. For instance, it includes shortcuts for your most-visited sites. (You can’t edit these as you can on the standard Edge’s new tab page, but you can turn them off.)
Another difference: If you enter a URL in the text box on the Copilot Mode's new tab page, the browser takes you directly to said page, rather than to a chatbot response or a search result page. I ran into some weirdness at one point, however. Copilot showed me a message saying it wouldn’t open a web page because it was either blocked by a login or “racy.” I was asking to open the Copilot Daily podcast, so go figure.
(Credit: Microsoft/PCMag)I initially missed the engaging photos on the regular Edge new tab page, but you can enable those in the settings without leaving Copilot Mode. You can also still use an alternative default search site, such as DuckDuckGo.
(Credit: Microsoft/PCMag)One major difference between Copilot Mode's new tab page and that of Copilot itself pertains to the drop-down menu that's accessible by clicking the Copilot logo to the left of the text box (above). On the Copilot Mode page, you see options for (in order) Search & Chat, Ask Copilot, and Think Deeper. You might also see links to beta Copilot Labs features like Actions. Use Search & Chat if you want Copilot to take the web page you’re on into account. Choose Ask Copilot to make standard Copilot informational and text or image generation requests. Use Think Deeper for projects that require more research, such as choosing a car to buy. That last option relies on OpenAI’s advanced o1 AI model and can take up to 30 seconds to return a result.
When you opt for voice interactions, the choices are different. You can’t generate images and text using voice prompts, for example. Once you switch back to a text interaction, however, the browser takes you to the regular Copilot web interface, meaning you lose the Search & Chat and Action options. It’s an inconsistency that Microsoft needs to iron out; the new tab page should always offer the same options. In any case, the standard Copilot options are Quick Response, Think Deeper, and Deep Research. The last option handles even larger projects than Think Deeper. It can take up to 10 minutes and generate several pages of information.
Keep in mind that you don't see text responses for voice interactions with Copilot Mode. The regular version of Copilot shows you a text transcript for voice interactions after you turn off the mic, which I prefer.
Quick Assist
Hit the Copilot button to the left of the address bar atop Edge to open the Quick Assist panel, a compact version of the Copilot interface. This feature is available no matter what’s in your main browser window. From this box, you can ask questions about or get summaries of the current site or pages across multiple tabs (see the next section). When I asked Copilot to summarize a page about a conservation project, it gave me a rather long bullet list of key points. I then asked it to summarize the information in one paragraph, and it complied, with an informative (though long) paragraph. And when I asked when the project took place, it gave me the correct answer.
(Credit: Microsoft/PCMag)An example that Microsoft gives on its Copilot support pages is asking the AI to “skip the life story and get to the recipe.” This ability is useful beyond recipe web pages. Plenty of tech tip articles, for example, start with a long diatribe about the history of the feature or problem, which you might want to skip to get to the tip steps.
Voice-Driven Browsing
While in voice mode, you can open web pages without lifting a finger, and it’s cool to see Copilot navigate for you. In testing, I asked the AI to find the best price on a Samsung soundbar. Copilot opened the Walmart webpage for Samsung soundbars. I then asked Copilot to compare the prices with Amazon’s, and it opened the relevant page.
I was also wowed when Copilot quickly scrolled the page down to the correct item after I asked it to find the least expensive soundbar on the current Amazon page. Copilot Vision, conversely, only sees what’s visible on screen. Google’s Project Mariner offers a similar capability, though you have to pay $249 for the privilege of trying that not-ready-for-prime-time feature.
Copilot Mode also has translation and unit conversion capabilities. I asked for a translation of a page about the Mexican environmental agency to test the former. My Spanish is fair enough to know that the translation was spot on. Here’s the interaction:
Actions
These are perhaps the best examples of the future of AI browsing, though the current implementation in Copilot Mode falls short of the idea of completely hands-off agency. Actions use third-party web services to complete tasks for you. You need to choose the Actions mode from the search box’s drop-down menu on Copilot Mode’s new tab page to get started. I tested the capability by asking it to make dinner reservations for two near Lincoln Center tonight. Copilot returned a page within a page, like this:
(Credit: Microsoft/PCMag)You can either let Copilot drive by pressing the play arrow or take control by clicking, you guessed it, Take Control. I asked it to find a cheaper option that’s still highly rated. This took Copilot some time. In the interim, it displayed a 'Thinking' message with notes on its process, such as “scrolling down for more restaurants.”
(Credit: Microsoft/PCMag)Finally, it found a table at Rosa Mexicana and asked me to enter my phone number to confirm the reservation.
(Credit: Microsoft/PCMag)I didn’t actually make the reservation, but I was satisfied that the AI got me to that point. I’m not sure the process was much more efficient than going to the site myself, but it’s a decent proof of concept. I did, however, ask it to remind me about it in two hours, and it agreed. This surprised me since when I'd previously asked Copilot to set a timer for me, it said that that capability wasn’t in its repertoire.
Multi-Tab Analysis
You can ask Copilot about all your open tabs at once. The Quick Assist panel is handy for this. One use case could be for a price comparison across sites. I tried the feature by opening four websites’ browser comparison articles and asked which browsers won, based on all the open tabs. The panel displayed a table summarizing the information from the pages across categories, such as performance and privacy. Again, this is not something you can do with Copilot Vision, which sees only what's visible on the current screen.
(Credit: Microsoft/PCMag)What Copilot Mode in Edge Can’t Yet Do
You can't use your voice to have the AI generate images as you can with the standard version of Copilot on the web. And unlike earlier-generation AI assistants, it fails to set a timer for things like cooking food. It can’t take you to a specific moment in a video either, as Comet can, thanks to its Google integration. Copilot Mode also can’t point to and highlight items on the page like Copilot Vision in Windows. I also lament that one particular voice command doesn’t work: “Stop listening.” This is essential if you’re using voice-only interactions.
I’m also still waiting for several announced features, including “Topic-Based Journeys” and “pick up where you left off” for multi-session projects, such as planning a trip. Microsoft claims that agents in Copilot Mode will be able to anticipate your needs and take actions for you, but the current iteration of the Actions feature hasn’t yet reached that stage.
A Promising Work in Progress
Microsoft’s blog post on Copilot mode states, “This is just the beginning of our journey in introducing new AI innovation into your everyday browsing. Copilot Mode is experimental and will evolve over time.” My testing bears this out. The tool is lacking in several ways to the point that it might not be worth switching to unless you’re simply keen on testing new technology. It stills lacks standard Copilot text- and image-generation capabilities in voice mode, along with basic AI features like the ability to set timers and a “stop listening” command. It didn’t always open the web page I asked for, either. And it can’t click on links on web pages.
A lot of Copilot Mode’s most powerful features require privacy permissions, such as history, logins, and payments, too. Microsoft states that Edge will mark these permissions with “clear visual cues” for full opt-in. Hopefully, we won’t end up with another Recall-like outcry. The company encourages testers to chime in on Discord about their experiences and wishes.
What's the best thing about Copilot Mode in Edge? Anyone can try it now without installing a separate AI browser. Perplexity’s Comet browser is in a very limited closed beta (unless you fork over a $200-per-month subscription fee), and OpenAI’s attempts at a standalone web browser are still at the rumor stage at the time of writing. Although Copilot mode in Edge does have limitations, it’s ahead of the game already and definitely worth checking out.