UnplugIT – 10 Things to Know Before Deploying Copilot


In this episode, Stephen talks to new co-host Darrell Webster about the 10 most important things you should know before using Copilot in your organization.

💪 Thanks to this episode’s sponsor: PowerApps911



Hey, everybody. Welcome to UnplugIT. I’m your host, Stephen Rose. Thanks for joining us.

I’ve been on the road for the past few weeks. I was in Chicago at EdgeCon. I was then at the Tech Mentor Live 360 conference and talked to so many folks about Copilot. I did a session really on, how do you get the most out of Copilot? What is the unknown, unwritten user manual to set yourself up for success? Now, a few weeks ago, we had our Teams day at Petri. I did an interview and an AMA with Anupam Patnik, who is the Senior Product Marketing Manager for Teams, and we answered all your Teams questions. I did a session on Copilot and what it is and how to prepare for it. This is the other side of that. I mentioned it during the show that we would do this. How do you really internally prepare so that you really get your money’s worth? This is that session today. As I thought about it, I really wanted to dig into, who would be a good person to join me on this? Someone who’s technically savvy, someone who gets it, but someone who is actually now using Copilot on the other side as an end user, and what that experience looks like. I thought, “Hey, I’d like to bring back a guest for the first time, for a second time on the show,” and if that were going to be anybody, it would be my friend from the other side of the planet, Mr. Darrell Webster. Darrell, how are you, my friend?

I’m great, Stephen. It’s an honor to make the grade and come back to the show. You are part of the Two-Timers Club. Take a minute. I know you do a lot of great stuff. You’ve got Modern Work Mentor, your Darrell as the Service Twitter handle, you do a show with Daniel Glenn. Take a minute, fill the folks in on who you are, what you do, and how they can connect with you.

Yeah, thanks, Stephen. Look, as Darrell as a service, the angle that I’ve always taken when working in the community, working at roles and jobs is that I like to understand the tech, get to know how to use it, turn it on, see what is possible with doing it, but more importantly to me is helping people understand it and use it for themselves to align it to their business needs. So that’s my service. That’s what I do, and I’ve always taken people along that learning journey and the places that I create content, but also in the role that I take wherever I work.

So, you know, today I’m certainly still a change manager and working at Rapid Circle, and that’s where I’ve landed as a place where they’re using Copilot. And in the community space, I’m creating content at Modern Work Mentor on YouTube, and you’ll find some occasional blog posts when I get off my butt and write stuff at the same, modernworkmentor.com. Awesome. A lot of different things you’re doing, and you are so great. Your videos are awesome. You do a lot to give back at the community. You love folks to write you and ask you questions and engage. You’re super engaged, and I appreciate that because it really is all about, you know, what can we do to share the knowledge that we have or the knowledge of people that we know and share it out with everybody to help them do their job? That’s really all that’s in it. So before we jump into this, I want to take a quick break and go to a video from our sponsor. Then we’ll come back and we will jump into how you can start to get ready for Copilot. So let’s do it. Thanks, Stephen. Just a reminder to all of you, if you have Office 365, you have the Power Platform. Power Platform is a low-code, no-code platform that lets you build your own apps, workflows, reports, all without writing any of that hard code. If you want to learn more about it, you can go to training.powerapps911.com. We’ve got on-demand training, we’ve got live training, we’ve got private training, we even have a whole immersive university program, or heck, we’ll even do the project for you if you don’t want to get your hands too dirty. All right, back over to Stephen. – Okay, we’re back.

I found this cartoon online from a guy named Tom Fishburne. You can actually go to his site. He has great stuff. If you want to get one that you can use for PowerPoint, you can pay for it, I paid for this one. But this sums up the Copilot conversation I have with so many folks. How is this going to change things? I don’t know, but we need to have this like yesterday. Are you also seeing kind of the same sort of enthusiasm, but sort of just running in the dark with scissors kind of attitude towards ChatGPT and AI and Copilot and things like that? – I am, and I see the interesting thing is that it’s about accessibility to some of these technologies. You know, some of the free tools that are out there, of course, people get to experiment with it, try it out, use it alongside their workplace with or without blessings to do so. But yeah, everyone wants to sort of understand and get their hands on this thing that’s supposed to make work easier. – Yeah, and I was reading, it’s like what, 60% to 70% of employees are using ChatGPT or a similar tool to help them at work. The problem is one in three are putting confidential information into ChatGPT. And not that ChatGPT keeps that, but it is affecting what’s called the LLM, the learning language model. And that’s the tool that learns how to question, how to ask what you’re asking and looks at the raw data and brings it back. And there was a story a few months ago that Amazon typed in the words to some internal code names for some products and it showed up as part of the LLM because it thought, oh, this is a product, I wanna learn this and I wanna add it. So what many companies are doing is they’re building their own interface, which then limits how much can go out to the ChatGPT client, et cetera. So there’s a lot of back and forth. Have you seen folks who are now struggling with that and trying to figure out their way to sort of find that middle ground?

– Yes, yeah. And I think different companies are taking an approach to providing that sidebar. And I think Microsoft’s example of the browser edge and having it as a sidebar to have those conversations. And I know you’re gonna get into this a bit later.

It is a challenge, it is a challenge to make sure that you’re not sharing more than you should, but then the goal of the tool is give me something and I’ll help you reword it. Well, we’ve got to put it somewhere. – Right. And the problem is there’s also, there’s so many Copilots. There’s the Windows 11 Copilot, which I hear is now gonna come to Windows 10. There is the Bing chat and Bing chat enterprise, which is also now just being known as Copilot. There is Copilot, Copilot, there’s sales Copilot. There’s now a Copilot in coming to OneDrive. There’s one in SharePoint Premium.

It’s just insane how many different Copilots are coming and which ones are just putting information straight into ChatGPT and bringing it back, which ones are only looking at internal, which ones are encrypted, which ones aren’t. So Copilot is not just one thing. It is a lot of different things to a lot of different people. Now I’m not gonna go into what you need to do to prepare, like, hey, you need the new Outlook client, the new teams here and all of that. We’re gonna dig into a little bit more than that. So let’s go back to our slides and let’s take a look. And let’s talk about how Copilot works. Now, the first thing to understand is Copilot is not ChatGPT.

Windows 11 Copilot, ChatGPT. If you’re using Bing, that’s ChatGPT. You’re basically entering in information that’s going out to ChatGPT. You don’t need a subscription. You’re guaranteed to get your answers quickly without having to be in a wait list, et cetera. And that’s the benefit of doing it that way, but you’re not. But it’s still out in the wild. Even if you’re using Bing chat enterprise, now called Copilot, that’s encrypted, but you’re still using ChatGPT. The big difference with Copilot is two things. Number one, you’re only looking at information within your own company when you are searching for certain things. Yes, it can go out to the web and bring in information, and it will do that using chat GPT, but anything you’re searching for, any of your company information, stays inside your company. So there’s three kind of key areas. First of all, all of your content has to be in the cloud. So we’re looking at Microsoft 365 apps. Those can be desktop, those can be web.

 It can be on a Mac on the web, et cetera, but that’s kind of the key area and all that storage has to be in OneDrive, SharePoint, Azure, et cetera. Now what happens is when you go to ask something, it really looks at two areas. It has something called Microsoft Graph. We have our Symantec Index and the Learning Language Model, LLM. LLM is what Microsoft has basically purchased or funded through OpenAI. It is that engine that they use for ChatGPT that they have picked up and brought in a private version that Microsoft folks who are using Copilot are using. It is using that engine to basically do three things. It connects to Microsoft Graph, which is where all of your content basically lives. That’s all your emails, files, meeting chats, calendars, contacts, everything that’s in the cloud across Exchange Online, all of those apps. And then you have something called the Symantec Index. The Symantec Index is what breaks down all that content into individual bits. So it looks at a two line in an email and says, “Ah, here are people that you sent an email to.” It looks at an attachment and says, “There was an attachment to this.” Or if you did a meeting, “These are the people in the meeting. Here’s the transcript, here’s the recording. Here’s the files that were shared in the meeting. Here are the action items and to-dos. Here’s who was not there in the meeting.” So Symantec Index basically tags all of that. So when you ask a question like, “What happened in the last meeting?” Copilot goes to the Large Language Model. The Large Language Model goes to the graph and pulls all the bits of data that are relevant

 and says, “Ah, you were in a meeting on this day. So here’s the other people who were in there. Here’s the action items. Here’s the files that were shared. Here is the transcript. Here are the points in the transcript where you were mentioned. Here are the things that other people have searched for and looked to find that are important and puts that all together and gives that back to you inside of Copilot.” And that is very, very simply how it works. It’s looking at your data, finding the data that’s relevant, tagging all that data and bringing it together. So that’s it. Darrell, your thoughts, or did I miss anything with this? I’m simplifying it a bit. And Jeremy Chapman’s got a great hour-long video on how this all works behind the scenes, but I don’t know, how would you rate that discussion of how the Copilot works? – I, yeah, I think that’s great. And if I can talk very briefly about seeing that action. Now I started a new job.

 Copilot of course is new to me. I’m not gonna pretend to be an expert in something I’ve just got access to,

 but it was really interesting. I wanted to use it to get to know my role and people as I onboarded over the first couple of weeks. And so going through that same flow just very quickly, it was like, “Okay, I’m gonna use M365 chat. I’m gonna ask about the person I’m about to meet soon, Lauriam, what’s he been up to? What sort of things do we have in common? What’s he been working on?” And so chat came back and I used all those signals from the graph to show, and only show things that of course I have access to, chats I have in common, group chats, documents, conversations and teams and email. But it came back and it summarized that and gave me all these little data points. And one of the really valuable ones was that he’d been working on something earlier that morning that was gonna be relevant to my conversation. And so I was able to go in, look at the reference point, open up the work and the document and boom, I already had some more context. – That’s great, yeah. Cause it’s also looking at things like, where does this person sit in the organization? Who do they report to? All of that information inside of Azure Active Directory, which is now Entra, because we need more new names for things. But yes, all of that gets pulled together and it does it. So perfect. All right, so let’s go to our next slide.

 So the first thing that people say is how much? And Microsoft announced that it was $30 per person per month. Now that’s just for Copilot, that doesn’t include Teams Premium or any of the other Copilots. It does include the Bing Enterprise, now Copilot, Copilot, the Copilot, Copilot, but that’s it. And that is a per person per month license.

 And I posted, I said, this seems kind of high and I didn’t put it in context. It sounds like a lot of money. And I was really thinking about all the things that have to happen before it’s gonna work the way you want it to. And my favorite is a Microsoft MVP said, “30 dollars a month per user, that’s only one dollar per day per user. That’s not much money at all.” And I started to think about that. And I’m like, actually it is when you really start to think about all of the things that you need to do to get it to work as advertised. And I’d like to jump into some of that. So any thoughts here before I move forward? – Yeah, there was one other thing that some organizations face too. The minimum license that you need to pair with Copilot. Yeah, you gotta uplift that. So there’s a cost involved there. – Right, which is 300. And what’s interesting is I was chatting with a guy in Chicago, he said, “We have 250 licenses. They would not let us do it.” They’re actually saying that the minimum licenses you have to have is you have to have 500 users to get 300 licenses. They do not want people buying licenses that will not be used at this time. So they are forcing that 300 and forcing that you have at least 300 active daily users, preferably closer to 400 to 500 before they’ll even sell it to you. So people that are like, “I’m willing to pay for licenses aren’t gonna use.” They don’t care.

 And it’s a lot of it is because of the Azure usage, but we’ll talk even more why that’s important and how you need, it really needs the company data. And if it’s not getting enough, it’s not gonna do its job. And that’s what we’ll jump into. So here are the things that you need to do before you launch Copilot. And the first thing you have to do is you have to manage and organize all of your data. So what do I mean by that? Well, you’ve got Teams, Exchange, SharePoint, OneDrive, all of this data that’s out there. Now think about it. If you have somebody that has 10 years worth of Exchange data, and we know people like this, where they store everything in Outlook.

 If you have 10 years worth of Outlook data and you ask a question, the way that your company worked and responded and looked at things in its position within the marketplace 10 years ago is very, very different than the way it is today. So you have newer emails that reflect the company’s views and products and older emails that reflects products that you may no longer offer and past views.

 The other thing that you run into is companies that have not been using shared files that have not been sending cloud files or links, but sending actual files. You may have within your network, 10 to 15 different, slightly different versions of the same file across all your cloud endpoints. And what’s gonna happen when Co-POD has to find something, it’s gonna look for a file and say, well, there are three different files that are similar, that were all sent out a week ago. I’m gonna pick the one that’s newest.

 And the one that’s newest may not be the one that actually has the accurate information. So look at it this way. If you have, let me move back. If you have this much data and you start asking Copilot questions, you’re gonna run into issues because that is a lot of data that it’s gotta parse through and learn and try to figure out what’s right and what’s wrong. Start with this much data, let’s say only the work you’ve done in the past three years, everything else you archived, et cetera, then it’s gonna be much easier as people ask questions for it to start to thin that data and finally get to a point where it is a limited subset of data that is correct and valuable and is worked through all the old data that isn’t there. So if you have got tons of duplicate data, old data, things that are like that, it’s gonna take Copilot a long time to get you to accurate answers and accurate information.

 – If I got something to add there, yeah.

 And this is just thinking outside of the box, but for years people have learned to optimize search engine results and write things to make sure that results will be accurate and in the public sphere, you’re incentivized because you want your content to be there so you can get paid or get work or whatever. Internally, that’s gonna be interesting. We don’t know yet what the Copilot or the LLM is really looking for to prioritize. Do we need to now learn how best to create our content so that it’s going to be rated higher and we can make sure the right information is landing for people? – Yeah, and that’s gonna be syntax. Syntax will learn and grow and you can tweak it, but it’s gonna be really hard if you’ve got an average of eight to nine years worth of data and you have 10,000 users, that’s a lot of data it’s gotta go through. And what that means is it’s gonna spit out wrong answers and find things that are wrong. So my recommendation is anything that is older than three years, put it into Azure cold storage. You don’t need to delete it, but you need to clean out that email box. Anything that is older than 2020, it goes. Any files that are older than that, what is it?

 The average user only uses, what was it?

 80% of files that an average user gathers or looks at were created within the past 90 days. Out of the 20%, there’s generally almost nothing that they will view that is more than two to three years old, which means you’ve got a single digit percentage of files that are greater than five years. You need to keep them for compliance purposes or things along that line, but people are not accessing them. A, they’re taking up a ton of room, which you don’t need to pay for. And B, with technologies like this, it’s gonna make it much harder for it to bring accurate data.

 The next thing is team sprawl. Team sprawl is also gonna create issues. If you have too many teams that are similar and overlapping, and I’ve talked about this on other shows, you’re gonna have to reduce those down. Otherwise, again, too many things that are similar, it is gonna take it a lot longer to figure out where that right answer is. And then this one’s one of my favorites, and I know Darrell’s got some thoughts here, but you gotta train people how to use it. It’s not like Copilot sitting right next to you and says, “Hey, you should go to Copilot and type that in or ask this.” People don’t know what to ask and what to search for and what it’s capable of. That’s gonna take training. So you’re spending half a million dollars plus a year for this. You’ve gotta teach people how to use it and when to use it. Hey, when you’re in Outlook, here’s what it can do. When you’re in a Teams meeting and you walk in late, here’s what you wanna do. If you’re doing Word, here’s what you can, you’ve gotta teach this to people. They are not gonna figure it out much more than, “Hi, go get me information on this.” They’re not gonna be aware that I can say, “Hey, go out, take a look at these four decks and pull this together and do this.” So that’s training and that’s time and effort. And it’s also training people that not everything that Copilot brings up is always correct.

 Take a minute, Darrell, because I know you’ve done a lot of training and talk about what’s involved in teaching people to use something that they’ve never used before in a brand new way.

 – Well, there are layers to this and certainly it’s almost like its own episode, but keeping it brief and even going from my own experience,

 I’ve had to change the way that I think about searching, getting information, or what am I using this for? I’m not just going to Copilot because it’s a new thing and I’m trying it out. I have a need. So it starts with a need. And then when you go in to do that thing that you want to do, if it is about summarizing something to give a report to a manager,

 it’s then that you think, “Okay, I would normally do this myself, but what do I need help with? Do I need to summarize? Do I need to create? Do I need to analyze?”

 And so all of these sorts of things are becoming new patterns to try and use Copilot, which in one sense, it’s all about actually slowing down and thinking about what you want before you start. Because Copilot is all about giving a prompt. And if you haven’t sort of slowed down and thought about, “What do I want to tell this thing to go and do for me?” Then it’s a new learned behavior to slow down and think about that. – Right, and what it’s capable of, even things that’s not asking you. I had somebody that said, “Well, you know, if you’re using ChatGPT, it’s pretty easy.” And I said, “Have you ever rented a car?” He said, “Yeah.” I said, “So when you get in a rental car, let’s say it’s a really nice rental car, they upgraded you. Something you’ve never driven. How quickly are you able to pull out of that space?” He’s like, “Well, it takes a while. I got to figure out where all the buttons are at and all of this and how to connect my phone and is there Apple CarPlay in it?” And I’m like, “Aha, wait a minute. You said you know how to drive. So why is it taking so long?” And he’s like, “I see what you’re saying.” I’m like, “Yeah, it’s all different. You have to learn how this one works and when you can engage things and where things are at and when to use them and even new features in the car that you may not be aware of.” I remember the first time I rented a car that had ABS and I went to go stop and the wheels did that weird thing and I flipped out. I’m like, “What the hell is happening?” I didn’t know. I had to learn. So that’s kind of a really key part of it as you don’t know what you don’t know and you have to have people that are trained up, change champions, et cetera, to do that, which is gonna help you to get the value for what you’ve purchased.

 You have to change how people think about AI. Now, this is something I talked about in my session for Teams Day but people still are very wary of AI. They believe it’s gonna take their job away. They believe that AI is spying on them. I can’t tell you how many people I’ve had a conversation about Microsoft even the first time that they got that email saying, “Hi, you worked this many hours last week and you need more work-life balance time.” And you want me to put, they went, “Is my boss seeing this? Does my boss get a report of how many meetings and how much time online and what I spent?” And I’m like, “No, it’s anonymized data.” They only see it by manager or department. They don’t see individuals, but there are people who do not believe that. And unfortunately, because there are bosses who are looking to see when someone logs in and when their light in Teams changes, et cetera, it perpetuates that and creates a lot of issues. And that is a really big aspect of that. There are all thoughts on what you’ve heard from end users around AI that is not Terminator or Whopper, but still has that a bit of paranoia in it. – Yeah, yeah. It is certainly a level of trust. Can I trust that this thing that now tells me more about myself and my connections and the information available to me,

 is that also available to everyone else? And I think we do have hints, some places where we start to use Copilot, which actually tell you that. This is your personal experience of the meeting. You’re gonna ask some questions and get some answers back. And it’s going to be a level of trust to understand and get through. Is Copilot is my experience and this is how my information is handled.

 – No, I fully agree. I think that’s great. Here’s another thing. And you mentioned this earlier about content sharing, but here’s what people forget. Let’s say there was a folder for a year ago I gave you access to. And since then I’ve put other documents in there that perhaps are confidential.

 That will now show up as part of your Copilot. So you’re gonna have to take a look at what is being shared today and potentially turn off some shares, remove some shares or alter them. You’re gonna have to roll out a pilot program because you’re gonna have to teach some people how to use it, learn from them and adjust. And obviously this only works best if everything is in the cloud. So if you’re still doing some things on-prem, all of that has got to go to the cloud. Any thoughts on that, Darrell, before I go to my next slide?

 – Yeah, actually it is again, another big topic to dig into is shared and over shared content. And that’s why there are some things we’ll need to do to tidy that up and make sure that we’re only seeing what we need to see. – Right, and on the other end, if you go to a document and you don’t have access to it, SharePoint will come up and say, “Hey, you don’t have access to this.” That won’t happen in Copilot. Copilot’s not gonna say, “Well, there’s some documents you don’t have access to. You wanna get rights.” It’s just gonna say, “I can’t find anything and there’s nothing there.” So that also creates potentially a new level of complexity. You’ve got to manage and lock down all of your data across all your tenants. You’ve got to also see if there are things that require GDPR, data residency. I need to go multi-geo because this content can’t be seen by folks in the US, et cetera. You’re also gonna have to sort out the licensing structure. Who needs Copilot? Who needs Teams Premium? Who needs Dynamics? Who needs Viva Sales or CRM? It’s not just one Copilot license and not everybody’s gonna need it and some people may need several. There’s a feature in Teams Premium that when you walk into a meeting, you can say, “Catch up on what I’ve missed.” That’s Teams Premium. That’s not Copilot. And if you want that or the executive summary where it says, “Hey, here was the summary of the meeting. Here’s what was shared. Here’s your action items.”

That’s Teams Premium. That’s not Copilot. So there are different (indistinct) that are available in different products and that will continue to change. And those are extra fees. Teams Premium is an additional $12 a month. So now you have some people that are paying 42, some 57 if you want CRM Copilot but don’t have the ultra premium version, et cetera. So you’re gonna have to break all that down. And that becomes again more than just that $30 a month but it’s not even about the cost. It really is about making sure you’re gonna get your dollars worth. Any final thoughts on this before I move to our next topic? – Yeah, well, that’s where it is worth taking the time to develop some personas, understanding the roles and the needs of different ways that people work and aligning those licenses and capabilities. Some people are gonna find plenty of value in just having Teams Premium and not needing Copilot. – Right, yeah, exactly or vice versa. So yeah, you’ve got to really understand the capabilities, what it can do and who really needs what. Here’s some additional things to consider or as I like to say for those who are old enough, Danger, Will Robinson Danger, did they have that in New Zealand that they have? – Oh yeah, yeah, on a Saturday, reruns man. – Okay, thank you, Drew. I had some people who looked at me, they didn’t understand that and I made a P buddy in Sherman reference and they didn’t get that either from Sherman’s set away back machine. My God, you young people. Okay, good, I feel better. – All right, users still need to verify the accuracy and veracity of the content provided. The content provided may not be correct because it’s only as good as the data. The data is wrong or there’s too much data, may not be right so you can’t just trust it. There is issues with reinforcement of racial or gender bias used by the LLM and how it was changed. Again, the way you did business 10 years ago and looked at diversity, inclusion, equality, et cetera, is very, very different than the way you may do it today and you don’t want chat, GPL, Copilot leveraging those older models which do not stand up today. You can have incomplete bias wrong or corrupt suggestions and that will break the models and create issues and you have licensing, access to data ethics and how to deal with employees that are just gonna use Copilot to do a lot of their work. Is it wrong? Can you not pay them the same, et cetera. I’ll let you weigh in on this before I go to my next slide.

– Well, it kind of reinforces the thinking of you have to slow down to speed up. That when we’re thinking about information, we’re getting back, we have to slow down and say, is that really right? And it takes me back to an old maths teacher of mine. He said, don’t just take what the calculator says to you, take a quick look and just think, is that the right answer? We have to do that. So the benefits of Copilot helping us to speed up and find more time, we’ve also got to find that balance of just take a quick look and think, is that actually what it should be? Check out those references that are being provided.

– Exactly, perfect.

 There is the possibility of Copilot violating copyright laws because if you did a research paper on a,

 or competitive analysis it could bring in some of that stuff and you’re not aware of it. So again, knowing those sources are important. The quality of the training data used to train Copilot could lead to biased or incorrect suggestions. We talked about this. There is potential for Copilot to be used incorrectly and actually get into arguments with AI. I had something I was playing with one day and I asked for a date for something in World War II and it gave me 1872. And I’m like, no, that is absolutely not correct. It said, yes, it is. And here’s the references.

So nothing better than arguing with an AI that something’s wrong. You just have to go, it’s not right. I’ll ignore this and move on or read, put in that search. But it’s kind of terrifying. – It’s okay if it lands from it. – Right, but it wasn’t, and it was telling me I was wrong. I’m like, okay. – Yeah. – Data quality and quantity, which we spoke about. And then here’s the other thing. If you’re bringing 3P apps in, you’re bringing, let’s say into Teams, you’re bringing in Confluence or Cloud from Jira or Salesforce or SAP or any one of those. You’re now bringing in other data from other apps. Again, that may not have been through syntax. It may not be verified. So that’s going to add to the level of potential incorrect data or incorrect copilot statements. Any final thoughts on this?

– I think that’s, no, that covers it well. And I knew what you mean by having those arguments with copilot.

Again, slowing down, is it what it should be? Okay, and then just correcting it. Is copilot going to learn from what I’m telling it now to say that? I think the copyright one is important too. And it’s definitely a hot topic, especially in the public sphere, where creative content, artists, photographs, that sort of thing.

It’s important to recognize what that source is and give it credit where credit’s due if anything. – And it may even create something that is original, but is based off of something or really looks like something that is well recognizable.

And leveraging or using something like that, even though it is unique, is still inferring something very simple. If you use McDonald’s red, McDonald’s yellow, and you make an N, it’s gonna make people think, “Oh no, that’s McDonald’s,” because of the colors, perhaps the shape. And even though it’s not it, you are inferring things. So it can be, there are certain colors that are licensed, Pepsi owns, I’m sorry, Coca-Cola owns Coca-Cola red. That red color is actually a copyrighted,

Pantone color that they have the rights to and other people can’t use on products. I did their orange. So there’s a lot of things like that. So it is not that it is trying to cheat, you just need to be aware and again, do your due diligence.

– Well, yeah, Stephen on that. And this is just one for Copilot coming to Microsoft designer.

And the thought, well, this was announced of course, how it’s going to use your brand asset library. The same challenge is going to be faced internally. So imagine the designers and internal commerce people that have put a whole lot of effort into, this is how we want the image to look. And designer may take and combine things and they go completely off brand. – Right, in a very weird way. This was an email that got released out to the public. So I’m going to use it. It’s not something I got internally. This was on the verge, but it is, hey, they basically said, if you need to exercise caution, use it, due to privacy and security or things along that line, open AI, ChatGPT,

 we want you to be very cautious when using these tools. Now this is not Copilot, but it is still along those lines. And Microsoft’s made a serious investment in ChatGPT and OpenAI. This really goes a long way and it is good advice for anybody using any of these tools at any levels.

So I’m going to kind of start to wrap things up and then we’ll kind of go back and forth here a little bit, but you’ve got to do your research. You’ve got to figure out which, who’s going to need what, what do we need to do internally? You’re going to have to budget for this. And I will tell you, after doing these sessions and chatting with folks, there were people saying, yeah, we were going to roll out Copilot next week. I think we’re going to wait six months and get our house in order before we do it. We need to clean and migrate our data. We’ve got to create some change champions. We have to do some piloting and analysis on this to understand how it’s going to affect users and what we’re going to need to do to be able to change habits and processes to do this. We’ve got to think about how we’re going to secure all of this and look at compliance and governance. And that’s going to take us about six to eight months. That way, when we get Copilot, it’s going to work as expected. But if you rush in, buy those licenses today and just turn it on without taking these steps, it is not going to work as expected. Darrell.

 – First impressions are going to count. So whatever time you can put into preparing your environment and your people is going to help you get a better result. It’s really hard to win people back when they try something out and think, “Eh, no, no good. I’m just going to keep doing what I’m doing.” – Windows Vista.

Windows 8. Yes, exactly. They’re like, “Yeah, I tried it. I didn’t like it. I went back.” We had a lot of folks that went to Windows 7 and said, “Oh, I hate it. I went back to XP or Windows 10, et cetera.” But it is very hard. As someone who worked on OneDrive, when we completely relaunched it and rebuilt it, it was very hard to convince people to come back and try OneDrive, that it was different. And that’s a lot of work. What are we missing here? Or what didn’t we cover? Or what is out of all the things that we talked about, the one thing that you really want people to take away from them?

 – Hmm.

 I’m always going to be looking at the people side of the tick

 and there are some things that you can do early to explore and understand how things work. Try and leverage some of the free tools, the free experiences via, what’s it called again? Copilot free, no Bing chat for something.

 You can sign it. – You can have it in Windows 11 and you can have the Bing chat, which is part of Windows 11 or the browser, which is just pure ChatGPT. And you can leverage that. If you want the encrypted version, which is now copilot, you can buy that for $5 a month if you don’t already have an M365 enterprise license. So that way, whatever you’re sending back and forth is encrypted in case people are sending things that are research, company research that you don’t want other people to find or understand.

– Yeah, so early things that you can do to prepare your people, we are going to need to understand that it’s not how we’ve been searching the way we’ve always searched. If we can start to use those free tools, start to think about, I need some help with creating content, learning how to become proficient at prompting, and then you’ll have a better experience when you start to liven up copilot in your organization.

– I think as you dig into the organization and you start to find out who your change champions are and start to get that group going, it’s gonna be great to sit down with them and say, what are the three things that you do as part of your job in help desk, HR, marketing, whatever that is that are time consuming or painful or difficult? What are three things in general at work that are hard? You’re probably gonna get people saying taking time off and you can say, great, we’re gonna teach you five things that copilot can do when you come back from being off for a week later and make things easier. Hey, summarize my emails for the week, summarize the meetings, I missed all of that.

Automation, the same way that we’ve talked about where you could use Power BI or Power Platform, if you have invoicing issues and who has it and where is it at, by better understanding people’s jobs and the different things that are difficult about those jobs, that’s where you can introduce copilot in a few small areas. Yes, I have to pull a monthly report that comes from 12 different things. Great, I’m gonna show you can do that and copilot now in 10 minutes. You’re still gonna have to go back and check and make sure the numbers are good, but it’s gonna do that. Or I have to create a marketing presentation every month. Great, here’s how copilot can pull together the framework. I think at the highest level, what’s important is that copilot is not going to do your job for you, but it’s gonna do a lot of things that are time consuming and pull it together for you. Much like these fridges that can see what you have inside and it creates a grocery list on what you’re low on or if you put in recipes that says here’s what you have to buy at the store, very much the same way. Create a framework for a presentation on this. Go out, find the numbers and give it to me so I can create an email on this. Look for outliers in these spreadsheets or things I haven’t missed. And that’s where AI is great is pulling together things that are not difficult but time consuming or digging into something to find those hidden nuggets that you might have missed. And that’s where I think companies will find the largest benefit from ChatGPT, not using it for everything, but for these things, but you as a company bringing this in really need to understand what those things are so that you can prevent, not so that you can present Copilot, not as this big, gigantic thing. But hey, here’s a tool and here are three things that you can do with it today that are gonna help. And as people use that and love it, great, let’s get you more. Go ahead and start to play with it. Share tips and tricks with other people within your organization. It’s gonna be naturally growing with that and that’s gonna be the best way to do it. And I’m just gonna remind everybody what happened when you just rolled out teams to everybody and didn’t train them. You’re now three years into teams and they’re still using it for chat and calling. They’re not using it for collaboration. They’re not storing files in there. They’re using tons of group chats, but not using team chats because you didn’t sit down, understand how people were gonna use it, where it was gonna solve problems, change champions, set it up and train. You’re gonna have the same issue, except this time, it’s not a free product that’s part of your suite. It’s something that you’re paying $30, $40 a month per person on. And at the end of the year when your boss said, “How did it go?” And you spent a half a million to a million dollars, that’s not gonna be a great meeting when they go, “Yeah, nobody’s really using it.” And that’s why. And that could be your job. And we don’t wanna see anybody lose their job.

I don’t. Final thoughts, Darrell.

– I’m glad to see that some thought has been put into adoption guidance and assets as well. – Yeah, Cara Waughn has done a great job. adoption.microsoft.com slash copilot. Sorry, I just wanna throw that in. Please finish. – Yeah, exactly, yeah. That’s gonna be a good starting place for customers to take a look at and plan for it. But to your point, Stephen, it’s the way that we’re going to have to learn this, being that copilot is an individual experience, but it’s also gonna depend on us sharing what we’ve learned, what we’ve known, bringing that context together. So having a center of excellence, a community where you can go in and talk about what’s been working for you, what has worked well and sharing those stories is going to help you give or share that value very quickly.

I think, yeah, if anything, do think about slowing down to speed up. And Stephen, if I’d summarize what we’re talking about today, it is just step back for a bit, see what we can do to prepare. And then once you’re in there, you also need to slow down to think about what you’re asking to get a good result. – I agree, 100%. I think you said it beautifully. Awesome. If you have questions, wanna talk more about copilot, things like that, reach out to me. I’m Stephen L. Rose on Twitter, or you can reach out to me at Stephen.Rose at Petri.com.

Darrell, anything else that you wanna plug, promote, talk about? I found out that I’m gonna be at Experts Live Europe in Wiesbaden, Germany in May. So I’m excited to finally get back to Europe and go see friends. Are you speaking anywhere anytime soon? I know that you’ve just started up a new gig, but any place people can find you. Where will you be doing the comedy stylings and what comedy club will you be at next?

– Yeah, yeah. Actually, the next place that I’m talking at is Digital Workplace Conference in New Zealand. And it’s in Christchurch next year in February. So it’s local, but we’re bringing a number of international people down too, to speak to that. And really, maybe I just put it out there. I am willing to come to different places around the world if you want me. So yeah, feel free to reach out. – That is awesome. I’m also gonna do a quick plug that in next month, it may have already happened by the time this show comes out, but I am the MC for the World Esports Excel Finals in Las Vegas at the HyperX Arena. And that will be replayed on ESPN, Oto and on YouTube. And I’m excited because not being an athlete, I never thought I’d see myself on any ESPN and now I will. So there we go. That will be, but it is a super Uber nerdy thing. I’ve ordered a shirt that has all of the Excel error codes that I will wear at this event. And I’m gonna be with some of the nerdiest people on the planet and I could not be more excited for this honor and opportunity. – What does an action replay look like for Excel in a like a command? – I don’t think there’s a–

– We’re going back to this. It was an amazing move. – Yeah, they actually do background commentary while they’re doing this. So there’s a whole list of things they have to do. And like, look, Bob just used a macro to do this and Jane just did the shortcut that automatically format at that. It’s a whole series of things that they have to do. And it’s not who only gets to it first, but that’s a big part of it, but also shortcuts and things that they used and everybody can watch everybody do it live and there’s a timer and people, I think it’s 25 or $30,000 that people get for doing this and there’s a live audience watching. So yeah, it’s gonna be pretty cool. So I’ve never been to or watched any esports online. So I watched last year’s show, it was pretty cool and just something different. So extra nerdiness for me. Yeah, it will be. And when it comes on, I’ll let everybody know where to go watch it and all the really bad Excel puns that I am already working on to use during that.

So with that, I wanna thank everybody for watching. I am Stephen Rose for Unplug IT and my good friend, Darrell Webster. Say goodbye, Darrell. – Thanks for having me, Stephen. And I hope you’ve enjoyed the show and found it informative. – Absolutely, thanks everybody. We’ll see you soon on another episode of UnplugIT. (upbeat music)