Copilot State of the Union

It has been five months since I provided an overview of the various Copilot products Microsoft was working. That piece, "A Flock of Copilots " looks quite outdated today.

In my defense, Microsoft has changed multiple product names (some even in the past week) to attempt to rebrand all of their multiple large language model technologies to simply be known as "Copilot". While there remains work to be done, the "Copilot" brand is now more secure than before, making it easier for end users and enterprises to understand what it is they are paying for.

Let's use this time to go through all the different Copilot products, explain what they are called, and which ones are at general availability versus public or private preview.

And before we dive into the products, it is helpful to classify them into two general buckets: those where Copilot is viewed as simply a way to attach natural language to an existing set of configurations, and the ones where Copilot is performing truly generative AI, and adding value beyond a "chat" interface.

Copilot for Microsoft 365

Copilot for Microsoft 365 reached general availability in November. It's important to note that this Copilot combines all the Copilots present in the Microsoft 365 application suite (e.g. Copilot for PowerPoint, Copilot for Word, etc.) as well as a standalone chat interface that just got renamed to be simply "Copilot" and lives either on the web at copilot.microsoft.com or inside of Teams in a chat app. The experience on the chat app and in the individual Microsoft 365 apps is not always the same (e.g. if you ask in chat to build an image, it can use Microsoft Designer to create an AI piece of art, but if you ask for an image in PowerPoint, it will grab a stock image from the PowerPoint image library) so having first hand experience with both is key. And because each of the Microsoft 365 apps is iterating at a different pace, some apps (e.g. Excel) are still in preview when it comes to Copilot.

These apps, especially in forms inside of Microsoft Loop and Microsoft Whiteboard, are truly generative - and make it easy to create quick "first drafts" of content - saving lots of time in the process and helping produce a better end result. Because Copilot for Microsoft 365 has access to the Microsoft 365 graph, it is also easy to see the direction Microsoft is taking it - today, for instance, Copilot for Microsoft 365 can look at your schedule and tell you what is coming up, but it cannot, yet, move your schedule around. That's coming in the future though - and so folks who adopt Copilot for Microsoft 365 today are going to see increasingly powerful ways to run their business. The meeting summarization capabilities may be the single greatest "killer app" of all the Copilots today - once you start to use it you'll never want to go back to the old ways.

GitHub Copilot

GitHub Copilot reached general availability last year. Despite being early to launch, many folks still do not understand the product - it helps developers write better code, more effectively. It is not some sort of "code gatherer" that is running around the internet looking for code to rip-off - and Microsoft has bolstered their AI creds by providing a guarantee to all who use the product (or Azure OpenAI services) that they personally back code created by it. Because many folks continue to spread FUD about generative AI, Microsoft has created a Copilot Trust Center that is well worth sending interested individuals to, whether on your legal or security teams, or those of your client.

This puts software developers in an unusual position: some clients, wrongly, don't want generative AI used in their code (to which I would argue that Visual Studio's "Intellisense" is likely used everywhere already) because they fear the code will violate copyright. Although developers should continue to review every line of code they produce - the far more likely source of bad behavior is, of course, bad developers themselves not the tools they are using. Other clients, having visions of AI generated code, think that a project that took a year should now only take six months. This is absurd in the opposite direction: although using GitHub Copilot will make some tasks quicker - it will also, by its very nature, make the code developed better, containing less bugs, and more documentation, than before. Technical debt is hard to see for most organizations and tends to emerge when codebases are moved or modified sometimes years in the future. Similar to the meeting summarization in Copilot for Microsoft 365 - whereby every meeting will now have easily summarized notes at hand - that doesn't mean all meetings will be perfect or that you can have 50% less meetings! My conversations with many clients around these products is that you will have more effective developers and workers. Less time will be wasted, even if you don't see an huge reduction in overall time. You'll get more accomplished, with less make work.

Copilot for Sales

Copilot for Sales just reached GA earlier this year. Out of the box, it natively integrates into either Dynamics 365 or Salesforce instances - and surfaces information about a sales process in meetings and inside both Teams and Outlook.

Because it shows up in both Teams and Outlook, we see some customers express confusion over the difference between Copilot for Microsoft 365 and Copilot for Sales, especially because many of the earliest adopters participated in Microsoft program including licenses for both. The license for Copilot for Sales also includes a license for Copilot for Microsoft 365, so you can think of it easily as simply an "add-on" that allows your sellers to insert rich information into their sales pursuits and do less tool hopping to record their activities. Whether sending an email or participating in a meeting with a customer, Copilot for Sales will provide immediate value and allow you to perform less make work, helping your sales close more quickly.

Copilot for Service

Copilot for Service reached GA at the same time Copilot for Sales did. And much like that version, it also natively integrates into line-of-business apps for contact centers, including Salesforce, ServiceNow and Zendesk. (And soon, Dynamics 365 for Customer Service.)

Similar to GitHub Copilot, Copilot for Microsoft 365 and Copilot for Sales, Copilot for Service is a "turn it on" service - it doesn't require a lot of configuration and customization to begin seeing value from it. The real benefit to it, much like the other products is two-fold:

• Changing your work culture to use the tool effectively and replace old habits through a rigorous adoption and change management effort
• Identifying which software products in your organization don't integrate natively - and would benefit from a similar Copilot approach

That last point is critical: we have lots of organizations that view connecting their line of business apps into Copilot as the ultimate destination for their IT organization. While that is a laudable goal - if you forget to do the necessary adoption and change management (ACM) with the low hanging fruit that is out there - then you're missing the forest for the trees. Products such as Copilot Studio and OpenAI services in Azure are powerful tools to build chatbots and to leverage large language models: but focusing on customization over ACM is a mistake. Nothing is worse than purchasing Copilot and forgetting to get it deployed successfully to change your work culture.

Copilot Studio

Copilot Studio hit GA when it was announced last year. How could it so quickly? because it is the product formerly known as "Power Virtual Agents" - now supercharged with LLM support to enable custom chatbot creation without complex coding scenarios. Although PVAs were "low code" - the ability to have natural language baked into the product, and to simply point it at a store of information, makes Copilot Studio a natural fit for organizations that have already successfully adopted many of the 'out-of-the-box' Copilots and are looking for further improvements.

Azure OpenAI Service

Azure OpenAI service hit GA over a year ago - making them one of the most mature products in the generative AI space for Microsoft. With these services, software developers can build fully customized generative AI products - allowing the full range of creativity for how to solve a business challenge.

Preview Copilots

Let's run through the Copilots that are in public or private preview next:

Microsoft Security Copilot is still in private preview. We anticipate it moving to public preview later this year - and when it does - expect many security organizations to adopt it to speed up their security triage process and produce better outcomes.

Copilot for Power BI is in public preview and requires an admin to enable it for organizations that have adopted Microsoft Fabric - which GA in the fall of last year. We anticipate it will reach GA later this year. Copilot for Power BI is similar to the other Microsoft 365 Apps such as Copilot in Word, or Copilot in Excel: it will help you generate reports more quickly using natural language prompting.

Copilot for Viva is in preview and as mentioned above, is about enabling natural language in many of the Viva tools, such as Engage, Goals, Topics and Learning. The ability to use Copilot in Viva is going to be tied to a specific SKU - but this will be distinct from the other Copilots (e.g. purchasing Copilot for Microsoft 365 or Copilot for Sales isn't going to enable Copilot in Viva).

Copilot for Azure is in preview and allows natural language prompting to manage resources within Azure and to answer questions about your Azure environment.

Copilot for SharePoint is in preview (roadmap says it will be released in March ) and similar to Copilot for Azure, is mostly about using natural language prompts to manage your digital estate in SharePoint. You'll be able to use it to create pages more quickly and effectively at scale.

Copilot for Power Platform is in preview today and shows up in multiple ways: in Power Apps , in Power Automate and Power Pages . In each of these, natural language prompts are being used to more rapidly create apps, cloud flows and web pages.

And finally, as was just announce post article completion - Microsoft has introduced Copilot for Finance in public preview to help organizations manage their digital financial estate with the help of natural language.

The future

As more of Microsoft's AI powered Copilot products are released and move towards general availability - it will almost become table stakes to have a natural language interface to many of the admin and creation tools we use on a regular basis. Being able to use prompts properly, via tools such as Copilot Lab, are going to separate succesful information workers and developers from folks stuck in the past.

And that's good news for everyone: because we all don't like taking notes in meetings, or making a simple mistake in a common piece of code, or having to spend hours building a "template" we should've been able to pick off the shelf from a previous proposal. Sellers don't like recording things in CRM: let's help them too!

Looking for advice on where to start your generative AI journey? Come ask us at Cognizant: we're drinking our own champagne as usual, and have experience with all of Microsoft's Copilot products and helping our clients adopt them successfully.