- AI agents like Copilot enhance individual productivity and skills, allowing people to collaborate more effectively. Each person brings their skills plus those of the AI.
- AI collaboration can produce outcomes of much higher quality and value than human collaboration alone.
- AI agents reduce low-value transactional collaboration, freeing up time for more meaningful human collaboration.
- Analytics are critical to understanding where and how AI collaboration drives the most value so it can be focused on priority areas.
- Organizations need data-driven strategies to determine the right AI rollout approach as capabilities rapidly increase.
The Microsoft 365 copilot experience is coming to you on November 1 for general availability. And as it does, it brings with it a new set of experiences around Microsoft Copilot Labs. What Microsoft Copilot Labs represents is a prompt repository and a space that we can learn from one another and from industry experts around what kinds of prompts are people using to take full advantage of Microsoft Copilot.
Let's look at another angle of this, which is the way collaboration changes. We've talked about productivity gains, and we've talked about some of the quality metrics, and then the importance of the data underlying it. But another interesting finding that we've had is the way it affects collaboration.
There's this old analogy: One plus One equals three. The idea is if you have an hour and I have an hour, and we collaborate together, we'll produce something, on average, that's a better quality outcome than if you just had 2 hours, right? Or if I just had 2 hours. And that's the whole point of collaboration, right? The sum is greater than the whole departs.
Now, there are some risks there, right? Suppose we add too many people to that equation, the skills and expertise we're bringing to the table that allow us to improve the quality. In that case, those diminish in value, because we're just not going to maximize or utilize those skills and experiences. And so, there's a cost as you scale collaboration, and there's also a cost.
The second, you have more than one person collaborating because what ends up happening is miscommunication, you have communication overhead and more. Therefore, the reason everything is not collaborative in most organizations is because of those costs, right? Because it's not always the best outcome. And from an ROI perspective, its return on investment is not as great.
Interestingly, in the copilot experience, we're seeing two major shifts. The first one is that each individual has an agent or an AI, like a Copilot. And so each person who's working with that, of course, is bringing their own experiences and skills.
But remember, they're also bringing all the skills that the AI agent has and the knowledge the AI agent has access to. In that scenario, I can do far more in a collaborative flow than I could ever do individually, especially when we talk about digital skills and how these tools give us more capabilities.
What's more is each person has their own Copilot. You might use it in different ways in the flow of collaboration.Let's say right here we're preparing an outline for an upcoming meeting. We're providing additional details, we're adding some additional information to it, we're turning it into an agenda. And each of those sequences, we're each using AI, but we're using it in different ways, right?
The other participant in this dialogue is using it to add content and pull content from various data sources, whereas here we're using it to do a lot of formatting a paragraph and then agenda work as an example. And so that sort of model of how we use it is important because it's both how we learn from one another.
The way people learn together through this collaboration flow is really powerful learning from these pilots, with these previews, and with Copilot. But it also changes again that value of outcome, because now we're able to produce more. One times AI plus one times AI equals maybe a lot more than three.
The other thing that's interesting is that the whole collaboration flow we just did can also be summarized, can be actioned upon, right, the data from transcripts to what activities happened, to even getting a sense of taking this result and combining with another result of a different set of collaboration.
All is wrapped around AI as well, for use cases and potential. When we see that again, 60% of baseline copilot adds more plugins, add more. You take something like this effect on collaboration, and it adds a compounding amount more that's already really interesting and worth thinking about.
One of the interesting and early findings we've had is that there's a lot of this low value collaboration that happens in organizations where it's kind of transactional. I say I don't have the digital fitness skills, I'm not the best at searching across sharepoint or tools like that. So I'm going to reach out to you, and I'm going to ask you to help me find something.
Now, that's a form of collaboration, but it's kind of transactional, right? It's low value. It's not like we're both using our experiences and skills. I'm taking advantage of your skills and experiences because I have a deficiency in my own. And this isn't the only scenario like this.
This happens with like I need your help compiling a PowerPoint presentation, I need your help doing X and Y. There are a lot of other skills where this comes up. And one of the things we've noticed with these copilot experiences is we've seen people shift some of that low value collaboration into copilot experiences. You use Copilot to do some of those actions for you. Semantic search to get the content I need, helping me craft and format a PowerPoint presentation. I don't need somebody else's skills for that.
Those are great starting points. But when you scale that a little bit further, you start to see that what it does is it leaves more time and more focus on high-value collaboration, where brainstorming, where humans have more experiences and skills, have more room to flourish and have an impact. And so that's a really good thing, I think, for organizations, because we have these big obstacles to productivity and most of them revolve around meetings.
In our data and analysis, we found most of those inefficient meetings and having too many meetings, et cetera, is actually stemming from this low value collaboration. The more we can offset that, the better. This also brings up why it's so important to think about numbers and data here. We're talking about analytics for a lot of the rest of the day, and we talk about analytics.
It's really important to understand that without being able to look at this from the right lens, you might just settle on, oh, it's great and people can use it. We might be missing the fact that it has other consequences as people use these tools, as adoption climbs, that we should tune and augment and nudge.
There are some groups that maybe really benefit from this who have high digital debt or who have a lot of that low value collaboration going on, that maybe those groups should be one of the first targets for copilot licensing or things of that nature.
So AI will lead to higher productivity and quality, but prioritizing the data quality, prioritizing security, sensitivity, compliance. Right, the right uses, which uses should we focus on first? Because there's just too much, there's too many opportunities, and this is only going to increase over the coming months.
As more and more features and capabilities come to market and more organizations like Microsoft and others learn how to use Copilot within the flow of work, it's going to make it harder and harder for us to keep up. So we need to make sure we're starting to build those muscles early to figure out what the right priorities are and what the right rollout strategy is.
