Mustafa Suleyman, a co-founder of Inflection AI and DeepMind (now Google DeepMind), was named CEO of Microsoft‘s AI group in March 2024. The Copilot experience has significantly decreased since then. Nevertheless, Microsoft has finally upgraded Copilot with a new look after six months under Suleyman’s direction.

Compared to the prior user interface, which adhered to Microsoft’s conventional design language, it represents a substantial change. The homepage of the new Copilot features excellent graphics and a card-based user interface with rounded corners. During my limited use, Copilot had a little more intimate feel. It’s warm and welcoming.

If you’re wondering where you’ve seen this interface before, Suleyman’s former startup, Inflection AI, offers a Pi chatbot (visit) with an incredibly similar look. The new Copilot shares the same focus on personal AI as the Pi chatbot.

As with ChatGPT Advanced Voice Mode, you can now voice chat using Copilot Voice, which offers natural interaction. I attempted to communicate with Copilot Voice, but there is currently a lot of traffic on it, so that’s one thing.

Subsequently, Microsoft introduced Copilot Vision, enabling online browsing in addition to Copilot Vision. It has the ability to see what is on your screen, and you may communicate with it to get ideas, suggestions, and more. According to Microsoft, this is an experimental function that will only be accessible to Copilot Pro users through Copilot Labs.

Copilot Vision is coming to the Edge browser, however it won’t function on all websites at first. Additionally, users will be able to choose whether or not to share their screen with Copilot Vision through an opt-in function. In addition, Microsoft claims that Copilot Vision will handle all data locally on the device in the wake of the Windows Recall debacle.

Microsoft has also included “Copilot Daily,” an AI-powered news presenter, in addition to that. On Copilot’s homepage, you can essentially listen to news from all around the world in an AI-generated voice. The four AI voices that are available are Wave, Meadow, Grove, and Canyon.

It’s difficult to identify which OpenAI model powers the new Microsoft Copilot at this time, but in my limited testing, it impressively solved every sophisticated reasoning question I threw at it. In contrast to OpenAI o1 models, it “thinks” quickly and correctly on its first attempt. I’m really impressed with the AI model that powers the latest Copilot.

Microsoft adds that as part of Copilot Labs, Copilot will soon receive an experimental feature called “Think Deeper.” Similar to OpenAI o1 models, it will require some time to “think” in order to answer complicated questions. It will only be available to Copilot Pro members who have paid for it, much like Copilot Vision.

So, have you had a chance to use the updated Copilot? Please share your thoughts with us once you download the Copilot app for iOS or Android or visit copilot.microsoft.com.