TECHnalysis Research Blog

September 22, 2023
Microsoft Copilot Updates Push GenAI to the Mainstream

By Bob O'Donnell

Ever since the company’s first AI event back in February of this year, Microsoft has been eager to position itself at the very tip of the Generative AI spear. Starting with its efforts around Bing Search through early glimpses of the technology being integrated into the Office suite and then Windows itself, Microsoft is clearly working quickly to bring the power of GenAI to as many of its applications and services as it can.

Up until now, the problem has been that many of these efforts have been focused on the individual capabilities of a specific application and not organized into a unified strategy. At its recent AI event in NYC, however, the company made it clear that it is now organizing these efforts around the Copilot name, newly designed icon, and more coherent vision. In addition, it is working to make the user interface of its GenAI-powered Copilot experiences more consistent across its range of products.

That’s a big step forward versus the more scattershot approach that different groups in the company had started to take. The net result is—or at least will be, when some of the new efforts are completed—a more comprehensive set of capabilities packaged and operated in a more consistent manner.

The two biggest announcements around the event are the launch of a new GenAI-enabled version of Windows 11 coming out next week, as well as the general launch of the Copilot-powered Office suite of applications in Microsoft 365 coming November 1. The Windows updates include a number of capabilities for controlling system settings via either a chatbot-like text interface or your voice, as well as new GenAI-powered versions of image and video utility apps.

The new Paint, for example, includes capabilities more associated with Adobe-type content creation capabilities such as support for layers and background removal. The updated Photos app incorporates easy background blur and things like easy object removal. The ClipChamp video editor integrates the ability to automatically edit raw footage down to a concise visual summary.

The new version of Windows 11—technically called Windows 11 22H2 update—also incorporates a dramatically improved email client that will be called Outlook for Windows (though it’s built from a different code base than the Outlook incorporated into Microsoft 365). As with its bigger brethren, it will integrate the ability to generate emails from simple prompts via a chat-based experience. A new version of File Explorer also integrates AI to more easily discover files without needing to know the exact title of the document or photo but just some characteristics about it.

On the office productivity side, Microsoft 365 Copilot includes all the GenAI-powered features originally previewed back in the spring and more. Importantly, the experience of using it across the productivity suite will be via a consistent sidebar style chat experience that Microsoft calls Microsoft 365 Chat. In addition, M365 Copilot provides access to all your documents, emails, Teams meeting notes and transcripts, and more, so that it can provide the type of informed digital assistant capabilities that made the initial preview of the technology so compelling.

What’s also interesting, and very important for organizations, is that it maintains all the data access capabilities that each individual has with the company. That means, for example, that Copilot for members of a team working on new product ideas could access all those documents as part of its “intelligence” but wouldn’t be able to see spreadsheets from finance department personnel, and vice versa.

Some of the many new capabilities that are coming to the general launch of Microsoft Copilot 365 include the ability to summarize an email thread in Outlook of both contents and participants. For those who often find themselves included in long email threads, this should make the process of getting caught up on your inbox much easier. Even more interesting is a feature in Outlook called “Sound like me”, which leverages access to your previous emails and documents to create a ghost writer-like feature that can emulate your style and tone when generating new emails. Integrating similar features in Word could have an even more dramatic impact on those for whom writing is a critical part of their job.

Speaking of which, the new GenAI-powered version of Word now includes document summarization capabilities and the ability to rewrite existing copy into one of several different styles. OneNote is also adding features for summarization and automatic copy editing and now offers advice on the pros and cons of different process approaches.

There are also some new apps being added to the suite such as Microsoft Designer, which leverages OpenAI’s Dall-E 3 model to do image generation. In addition to functioning on its own, Designer can work within other applications, such as Word, making it easier to create attractive documents.

Microsoft also announced the general availability of Bing Chat Enterprise for mobile devices. As with the desktop version, Bing Chat Enterprise is free to all existing M365 customers—unlike Copilot for M365 for which Microsoft has announced pricing of $30 per user per month. Bing Chat Enterprise assures the privacy of data entered into the prompt so that companies can use it for search purposes without data leakage concerns, but it doesn’t have the ties or integration with Microsoft 365 applications that Microsoft 365 Chat does.

To finish the event off, Microsoft also debuted several new Surface PC devices, including the Surface Laptop Go 3, the Surface Go 4 and, most intriguingly, the Surface Laptop Studio 2. The Laptop Studio has been upgraded to include not only Intel 13th Gen Core processors, but an Nvidia 4050/4060 or Ada 2000 GPU, and an Intel Movidius VPU/NPU, along with up to 64 GB of RAM, up to 2 TB of storage, and an accessibility-friendly haptic touchpad. Those are great raw specs on their own, but as Microsoft demonstrated at the event, it also lets this notebook run some GenAI foundation models directly on the device.

As Intel did earlier in the week with its Core Ultra (see “Intel Hopes to Reinvent the PC with Core Ultra SOC” for more), Microsoft showed Meta’s Llama 2 LLM (Large Language Model) running in a disconnected mode directly on the Surface Laptop Studio 2. While the effort was initially presumed to be done by the Intel VPU/NPU—which happens to be an earlier version of the NPU that will be incorporated into Core Ultra SOCs when they’re released on December 14—it was actually being run on the Nvidia GPU. This illustrates that real-world GenAI on the PC is starting to be accomplished in several different ways, all of which offer some potentially huge benefits in privacy and security. It also makes the case clear that client device based GenAI is coming a lot faster than many expected.

What Microsoft also made clear at this event is that it is continuing to move forward on GenAI-based experiences in an extremely aggressive way. The company sees its Copilot work as the computing paradigm of the future. To act on that vision, Microsoft will be quickly bringing it to the billions of people who currently use Windows 11 and/or the Microsoft productivity apps found in Microsoft 365. The impact of these moves is bound to be enormous, and the ripple effect of these launches will likely be felt for many years to come.

Bob O’Donnell is the president and chief analyst of TECHnalysis Research, LLC a market research firm that provides strategic consulting and market research services to the technology industry and professional financial community. You can follow him on LinkedIn at Bob O’Donnell or on Twitter @bobodtech.