Microsoft 365 Copilot Complete Tutorial: How to Get Real Value

Hello and welcome to this complete tutorial on Microsoft 365 Copilot. My goal here, and stick with me all the way

through and we're going to get to this goal together, is that by the end of this video, you will understand not just

what this is and how to use it, but how to get value from it. So, whether you're someone who's just been given a license,

someone who's deciding what to do in your organization, or someone who's had it for a while and is feeling a bit meh,

then there's something here for you. Let's get into it. We're going to start in the hero part of

the experience, which is the main chat experience. Now, you'll see as we go through that I'm going to be spending

some time here, but we'll also be looking at using Copilot inside some of the apps that you use every day cuz it

does follow you around where you are. But, most of the action happens here. So, to get started, open your browser of

choice. I'm using Edge and go to this website, m365.cloud.microsoft. Alternatively, if you have a M365

Copilot icon on your toolbar or a Copilot button on your laptop, you can press that and that will launch an app

that you can use. The website and the app are exactly the same experience. This is all just down to personal

preference. Now, we are working here with the paid, the full premium Microsoft 365 Copilot license. This is a

work license. If you're a consumer, you can't get this one. This is added on to your work license and you or somebody in

your organization, more likely, has paid this US $30 per user per month. And you'll know you're in the right place

because you'll see when you go to this website that you've got this work and web tab at the top here and also that if

you go down to the bottom here, you'll see M365 Copilot Premium. If you're not seeing that, if you're seeing Copilot

chat, go check out this other video. You don't have the stuff that I'm showing you here.

Now, the most important thing to understand about this, and this is where all of the value of the paid license

comes from, is that this is deeply connected into everything that you're doing at work in your Microsoft 365

world of work. It has got access to and understands everything you're doing in your email, everything that's going on

in Teams, everything in the documents in SharePoint and OneDrive that you're using that you have access to as a

logged-in user, as well as the context of your workplace. So, all of the organizational hierarchy, who your boss

is, who you're working with, who your collaborators are, who's in your team. All of that collectively is known as

Work IQ, and this is the core value. And you'll see what happens, I'm going to show you some examples of how this

works, but essentially what we're doing here is bringing AI into your work. So, this is very different from using and

most of you in your organizations probably either have some special arrangement with Chat GPT, or you've

been told not to use it for work. That is a public product. If you go putting commercially sensitive things in there,

that can be used to train language models or leak out, you should not be doing that. What we've got here is

something, this little icon at the top here shows you that this has got enterprise data protection. So, as a

start, whatever you're doing here stays safe within the tenant boundary, within the world of your work. So, this is not

being used to train a language model. Microsoft do not have any visibility of what you're doing unless you explicitly

say, "Yes, please share this prompt." And it does not go anywhere outside of your work. So, you are bringing AI

securely into where you work. And this is working within all of that Microsoft 365 access that you have as the

logged-in in I have access through Copilot to everything that I would normally have

access to even without it, but I can just do a whole lot more with it than I used to be able to.

So, this core work and web tab at the top here, we're going to start in work because that's where the value is. So,

that means when I come here and put a prompt in and ask Copilot to do something, it is by default using all of

the information that I have available to me at work as a logged in user. I'm going to start here with a very bad

prompt. I'm going to come back and talk to you about good prompting, but I just want you to understand this kind of work

grounding to start with. That's the way we talk about it is grounding. So, we are in the scenario that I'm

going to work through in this tutorial organizing a Pets at Work Day just because that is fun. So, what I'm going

to do to start with is just put a very simple prompt in here that says, "Tell me about a Pets at Work Day." Now, as I

said, this is not a good prompt. One of the core things to understand here is that this is not a search

engine, it's not just a search engine. As we go through here, we're going to be using it to help us do things, to create

things, as well as finding things. So, what we've got here, here's a clear view of Pets at Work Day grounded,

there's that word again, in what I can see in your organization and more broadly. This is a very nice

explanation. Thank you, Copilot, of exactly, it's almost like it's listening to me, of what I'm trying to explain to

you. So, when you're in the work tab, it is grounded in, has access to all of your information that you work with when

you're logged in with your Microsoft 365 work account, as well as more broadly out on the web. So, what happens is it's

bringing in different things around. So, something like often called this or that is just from its own general knowledge

or what it's found out there, and Wikipedia, so it's bringing in something about information about what's happening

sort of in the past about this common goals, how it typically runs. So, this is all quite general. But, what's

happening in your [clears throat] organization specifically is where it's picking up on

the stuff that we're doing. And because my prompt is very broad, where I've just gone, "Tell me about Pets at Work Day."

It's given me this very broad answer both external general stuff and internal stuff. So, you can see it's found a

document where there is actually a planning document because we are planning to do this. We've got

guidelines and it's talking about specific things. We've got a proposal. We've got some research. Uh there's an

email that's gone back and forth. So, you can see there evidence in this response that it's able to draw on all

of that context of my work. And there we have it. Now, this is going to show me the sources and you'll see we've got a

whole lot of stuff here. We've got citations from internal and external sources and it's gone quite quite far

and wide. So, that what that is what happens when I'm in my work tab. Now, we are going to do more useful valuable

things with this and write a better prompt. But, just to understand this work and web tab cuz it's important.

Let's say that I wanted to do something about Pets at Work Day, but I didn't want to reference all of my stuff at

work. So, I've switched across now into this web tab. And when I'm in the web tab, that basically goes,

"Ignore work for a second. Just give me a perspective without referring to what's going on here." This is still

secure. So, this is still enterprise data protection. But, now I'm temporarily saying, "Just don't look at

all of my stuff. I want something objective." So, you'll see the answer here is that this is giving you a very

generic answer. It's talking about Take Your Work Work Take Your Dog to Work Day is the last Friday of June. So, we've

got a whole lot of different things in here that are more generic. So, if I wanted to just kind of go out and find

information about that, that's useful. This is helpful as you're working with documents. Sometimes you just want a

fresh perspective on something without it kind of cycling back over things you've done. I've used this where I'm

for instance drafting a document and I'm trying to get it to give me fresh ideas and it just kind of keeps drawing the

ideas from the stuff I've already done. So, just be aware that you've got these two experiences where the high value of

this license the majority of the time you're going to be in that work tab because it's bringing all of that work

IQ into what you're doing, but it is useful to be able to switch that off temporarily. Just remember to switch

back again if you're not getting the results make sure you're back in the right place.

So, let's take a look at the concept of working with these documents and emails and things because you saw in the last

prompt that I just put in something very general and it brought back a lot of things, but often you want to help it to

find something more specific. This is a demo system. You can see I'm logged in as a fictional person called Miranda.

This is obviously not my real work, but I do use this in my work every day. I work in consulting. I have got lots and

lots of documents that are quite complex, multiple versions, and different people contributing to things.

And so, sometimes if I go in and say, you know, tell me what we've done with this project or tell me what's happening

with this customer, that's too much. Like if we've got years and years worth of history or things going on, I can go

in and point to specific things to give the prompt a specific source to say, this is the thing or these are the

things that I want you to refer to. So, the way that you can do this is let's put this same prompt in again just to

sort of give it a consistent experience. So, tell me about pets at Workday. Now, if I use I'm going to use the forward

slash key here. So, this is a the forward slash key on my keyboard that sits just at the bottom right near the

shift key. That's in my part of the world. If you are in a different part of the world, that key may be somewhere

else, but that's the one that you're looking for. And this brings up this context menu.

So, this is allowing you to find things and bring them in. The other way you can do this if you don't prefer to use that

key, I prefer to kind of use the keyboards keyboard shortcut, but the other thing you can use is this plus

button to add content. So, that will get you to the same place. You can either add work content, will take you back to

where we were, or upload, if you want to bring in something that's not already in the cloud, if you've just got something

downloaded to your desktop, you can upload it here, or you can browse your OneDrive.

Let's stick with where we were before, which is just to this context menu here. Now, you'll see that we've got filters

here. This is showing me all. So, what happens is that we have got all of the things that have been recently opened

here. So, I've got a pets at work day document, I've got a test notebook, a spreadsheet, an email address that's

been recently used by this user, but we can come in here and filter by different things. So, let's say I just want to see

people. This is giving me the people that I collaborate with the most, the files that I have most recently opened,

any meetings that I've been part of, including the transcripts, so you can actually use the transcript there,

emails that are going on in there, and more. Now, you can see here that if you wanted to access something quickly, if

you've recently been working with it, if it's recently opened, it's going to pop up in the list there. So, let's say

we're looking for this bring your pets to work day budget, for instance. If that wasn't there, then you can search

for it here. So, if you know what the file is called, you can say just start typing it in. Again, in a real world

scenario, rather than this lovely neat demo system, that can be quite hard to remember what it's called, or to have

something that where you don't have things called, like many things called the same name. So, the way that you're

more realistically going to work with that, if it's not in your recent files list, is to come in here. So, these are

the same two options we just saw in that plus menu before. I can upload from my desktop or from a device, or I can

browse cloud files, and this takes you somewhere where you can browse around and have a look in either your files or

in the SharePoint folders and sites that you have access to. So, we've got a folder here for Pets at Work Day, and

there's the Pets at Work Day budget, so I can select that. So, however I get to that, either I've just selected it from

the menu, typed it in, or found it, now what's happened is that I've got Tell me about Pets at Work Day, but I've asked

it to specifically refer to this budget document. Now, while it's doing that, let's just have a look over here. This

is a very It's a very basic budget, but you'll see this is actually focused on the cost of things. So, by referring to

that document, then we've got that context in there. So, now what we've got is What is Pets at Work Day? So, it's

given me all of those things again, and now it is bringing in the cost considerations because you'll see I've

referred to that particular spreadsheet. Now, I could improve this greatly because this is still a very poor

prompt, and I did promise you that we're going to we're going to talk about better prompts. But let's say I wanted

to come in here and sort of edit this. Tell me about our budget

for Pets at Work Day. Now, I'm asking it a far more specific question. Still not a good prompt. We

are going to improve on this either further even further. But now that I've kind of given it that budget document

and given it a more specific question, it is working with that. If I hadn't put that document in, it might have found it

in my demo system here where I've only got one nice neat document, it probably would have found it. But like I said, in

the real world environment where you might have multiple things and your naming conventions aren't so clear, the

AI might have trouble finding things. So, as a rule, using this clarity, telling it what source you want to work

with, and we are going to come back to this concept of a source in a moment, can help it do the job better for you.

Don't make it do the work of having to figure out which document you want to use as well as doing the job. Although

sometimes it is very good at just finding the things from a general prompt. So there's two different ways

that you can use that there. So this is the concept of it being grounded in your work. Now, while you've got this prompt

here, when I hovered over this, you saw that I went in and edited edited it. >> [laughter]

>> You've got another couple of options in here, but one that I want to come in and show you is this schedule this prompt.

So let's go down to the edited version because that was a much better one. Tell me about our budget for pets at work

day. If I come in here and schedule this, this might be something that maybe once a week I want an update on this

budget or I want to go in and check it again. Perhaps this is a work in progress. People are adding things in.

Again, I might write a better prompt. Please summarize pets at work day, the major costs in the different categories

or something more specific. But what I can do here is say, I want this to be, let's run it on a Monday and I want to

run it every Monday at

9:00 a.m. Got to love a 9:00 a.m. Oh, not 9:00 p.m. >> [laughter]

>> That's No one wants a budget at 9:00 p.m. on Monday. Maybe you do. 9:00 a.m. on Monday and receive an email and save

that. So what that will do is automatically run that prompt for me on a schedule so that I don't have to go in

and do it again. So to come back later and find those scheduled prompts, just go up to the three dots in the top right

here and select scheduled prompts. And that's where you'll find what you've created. Now you've got options here,

including forcing it to run now so that you can see the results immediately if you don't want to wait for the next

scheduled run. Just to show you what that looks like when it does run. If you've selected an email notification,

you'll get that. You can click through and that opens the Microsoft 365 Copilot main chat experience and you will see

the prompt run there and then you can just go ahead and continue chatting or working however you like with that.

Now, you've also got other options here. You can choose to edit, turn it off, or delete it entirely.

The other thing you might want to do is just save a prompt rather than running it on a schedule. So, to do that, if you

hover over a prompt that you like, select this little ribbon to save the prompt, give it a name that will make

sense for you if that's not it, and select save. And that goes into something called the Copilot prompt

gallery, but just to show you how to get to that, if you start a new chat, you'll see this list of suggested prompts below

the main prompt area. Expand that and that will actually give you a bunch of different options and access to this

prompt gallery with more things that you can try as suggested prompts and your prompts saved in that spot there. Let's

take a look at how to write an effective prompt because honestly, this is the biggest skill that you need to work with

any AI system. The rules I'm showing you here are for Microsoft 365 Copilot and other AI systems. If you think in this

way, you'll get the most out of it. So, rather than doing what I just did, tell me about the budget for this, tell me

about that, summarize this, uh find that, which is typically where we start with these very basic things. When you

put in a prompt that reads like that, you're essentially just giving Copilot a goal.

Tell me about this. This is my goal. But you're not giving it anything else to work with. So, I want you to think here

as if you were delegating a task to a person. If I just said to a junior person in my organization or to a

colleague in my organization, tell me about Pets at Work Day. What could happen? If it was someone in finance,

they probably would have told me about the budget. If it was someone in HR, they probably would have told me about

some of the logistics and rules and safety concerns. If it was someone who was the organizer, they would have told

me perhaps how many people had registered and what had gone on. So, you can see this is a very general goal, but

even if I talked to the humans with that little no context and no detail, then I'm not going to get very far. So, when

you're writing a prompt, I want you to think about this in the same way that you clearly communicate with your

colleagues at work, or as if you were delegating a task to someone else. And one of the most important things this is

This four things on this screen, but this is This is the one that makes difference, is the context. If you

explain to somebody why you want that, think about any time you've been delegated a task, or you delegated a

task to someone else, how much better is it? How much more sense can you make of it when you understand why you're doing

it. What are you asking Copilot to do? Tell me about the pets at work day, because I

want to understand if we're over budget or not. Okay, that's starting to get there. You can give it that context.

Tell me about pets at work day, and what the key things are that we need to organize so that I can send an email out

to the staff to ask for some more volunteers. That's completely different from are we over budget, from what are

our concerns. So, that context that goes around it is really, really important. The next thing is the source, which is

exactly what we've just been looking at. Which documents, emails, meetings, other kinds of examples do we want it to use?

Tell me about the Tell me about pets at work day, so that I understand the schedule of what's going on, and I can

communicate, and please refer to this planning document where the latest version of the schedule is. So, now

you're making it easier to delegate the task. Copilot probably could find that document if things are nice and recent

and clean, but making it more specific is going to make it less risky that it will give you the wrong information or

hallucinations. So, the more that you you don't make it do the work of finding the thing, you go, "This is what I want

you to work with, and here it is." And finally, the expectations. What kind of output do you want? Do you want an

email? Do you want a table? Do you want a 300-page essay? Don't ask for a 300-page essay. Do you want an elevator

pitch? Do you want a summary of the activities that are going on? So, if you consider all of these elements, you will

write a prompt that is far better than something where you just use the goal. I'm going to show you an example of what

this looks like, but you don't have to use every single one of these every time. And you don't have to do it in the

exact order, but I'm going to show you an example of what that looks like. The thing here is the thought process of the

understanding what other information should I be giving it so that it has the elements it needs to do a good job for

me. Let's take a look. Now, you should always start a new chat when you want to change the topic or reset the topic. It

just kind of clears the context, because you can keep doing follow-ups and editing in an existing chat, but it's

keeping the context of that thread of the conversation. So, when you're ready to kind of start fresh, make sure you

click that new chat button. So, here is my prompt. Tell me about the schedule for pets at Workday. That's my

goal. The next thing I'm going to do here is to say, "Including the timing and types of activities, so that I can

send this to our staff to get them excited about it." That is the context and a little bit more detail of what's

going on here. Can you see already how if you were being delegated this task, that extra information would help you do

more than just perhaps list the schedule? Then I'm going to say, "Refer to" and I'm going to use that forward

slash to bring up the document, and I've got this pets at work planning day, which is in there. So, I'm pointing it

directly to the place that I want it to refer to to get the schedule and then I'm going to add my expectation. I want

this to be an email that has a fun and friendly tone using puns for pets and animals wherever possible. Now I can

also put in a link. I could say use bullet points. I could say use a table, but I'm doing something a little bit fun

here because why not? All right. So now you can see it's confirming that it's referring to that document. Subject

we've got some cute little emojis here. Hello humans, it's time to unleash the fun because it's officially happening.

Now you'll see it's bringing in we've got the date from that document that's in there and the program.

[clears throat] This is so cute. 9:00 a.m. welcome session 9:30 and we've got all of those

things in there. Okay. So you'll see that that's very different from going back to this idea

of just a general goal if I said write an email about pets at work day that I just get what I get and that's it. So if

you're finding that you're not getting the most out of it, then the thing that will make the biggest difference here is

thinking about what's the source, what's the context, what are your expectations of being clear at it and then also

refining in that conversation if it's not quite right to give it feedback and ask for something a little bit

different. So this is a collaborative tool, right? We're in Microsoft 365. We've brought AI

into the context of our work and there's a whole bunch of collaboration tools that we can use here because most of us

don't work in isolation and so this is AI that supports that kind of collaboration. So let's come back into

this much more fun, friendly email and let's say I might want to work with a colleague to just kind of tidy this up a

bit or make sure we're getting the wording right. If you come down to the bottom here, you've got an option if

you're happy with it as it is, you can just select copy response and go and do whatever it is you like with that. We

will be coming back a little bit later in the video to have a look at how you can work directly in Outlook so that

you're reducing the amount of copy and paste. But, if you want to do something more collaborative and get someone to

work on this with you, you've actually got an option here that says edit in pages. So, if I select that icon, now

what I've got is a side-by-side, and you can change the the sort of width here. Let's go sort of 50/50 a bit more so

it's a bit clearer. Now what I've got is Copilot on the side and something that's called a page on the other side, and

this is completely editable. So, first of all, we can get rid of that sort of AI-based top and tail, and perhaps I

want to change some of these other things. Again, let's get rid of the the bottom here, and let's say this is from

uh Miranda and team. So, you can see that I've got this is like a live editable component, but you can share

this with others. So, I can come in here and say send this link or this component to a colleague, and that colleague can

be in here live editing it with me. Now, if you are familiar with Microsoft Loop, you'll have seen this

concept before. This is actually a Microsoft Loop component. So, this is one way that you can kind of work either

on your own or with a colleague side-by-side, and you can choose to take this out into a document, presentation,

or PDF, or add it to a notebook, which is something we're going to have a look at next. Once you're done, you can just

close that out, and you're back into the main Copilot experience. So, let's take a look at the concept of

a notebook, which is one of my favorite and most underused things in Microsoft 365 Copilot. So, if I come in here, and

we're going to say add a new notebook, the idea of a notebook is this is a place where I can very quickly collate a

bunch of resources together, bundle them up. Think of this as like a folder, like a workspace. And then when I want to

come in and ask questions or get summaries of things, it's just those documents are sort of all in there

together. So, when we're looking at the prompt before, I was saying use this document or you can add more than one

document or meetings and things in here. This is like just going, "Here's a folder of stuff that I'm working on

right now, and I just want to be able to chat to those things all the time." So, let's create a new notebook here. I'm

going to call this one Pets at Work Notebook. And in here, you can click on recently

opened documents, but you can also search. This is the same concept we saw earlier where you can search, you can

upload, you can link to something, or you can browse cloud files. So, we have got a planning document, I've got a

proposal, a benefits document, and a budget. There we go. Can't forget the budget. And then I just click create.

Now, I can go in here and give it some other instructions about we're planning a Pets at Work Day and some background

context. I'm just going to leave this as it is. So, what I've got now is this folder of all of these documents, and

now I can come in here and just, without any of that other context, say, "What is the schedule?"

So, what I've got now is all of those things put together, and this gives me a really quick shorthand. I've essentially

taken out the need for the context and the source in that prompting framework. So, writing prompts is really good if

you're just ad hoc doing things, whereas the notebook, I'm working fiercely on our Pets at Work Day constantly every

day. I want to come in here. I've gathered all that stuff together, and now I've got a shorthand between me and

this notebook that it knows the context of what I'm doing, and it knows the source of what I need to work on, and I

can just jump straight into it without having to do that. So, that's what's going on there. Let's come back in here.

We've got the same concept of start a new chat. The other thing that you've got in here is this get audio overview.

So, this is actually really good. We're going to go with a dialogue casual Ava. Let's go for a short one and

generate audio. So, this will give you a summary of everything that's going on here. So, this is perhaps I've come into

this project, I've gathered all of these things together, and I want to know what's going on. This can actually help

me with getting across it quite quickly. Now, this does at first seem like a little bit of a novelty, but especially

for people who are perhaps more auditory learners, or if you just want to be able to get across something quickly without

reading, this is actually a really high-value thing. So, let's take a listen to it.

>> Hey. Welcome to our chat about bringing pets into the workplace. We're diving into a proposal called Bring Your Pets

to Work Day: Enhancing Workplace Culture and Employee Well-being, organized by Miranda Lovell.

>> Oh, that sounds fun already. So, we're talking about pets, workplaces, and how combining those can boost morale.

>> Exactly. The whole idea is to see how pets can create a happier, healthier, and more engaging environment at work.

>> Let's skip forward a little bit beyond the intro into some more of the substance.

Oops, I've gone too far. Went right to the end. >> pet that can safely join the fun?

>> That's right. One big benefit is how pets help reduce stress and anxiety. When you pet an animal, it releases

oxytocin, which can >> There we go. We'll leave it We'll leave it in that happy place there. So, you'll

see we've got some options in here. I can actually come and save that to OneDrive. So, if I wanted to share that

audio file with other people, I could do that. And if I wanted to share this whole notebook with other people who I

work with as well, then we also have that option to invite them. They will also then get access to that content

that's been uploaded to the notebook, so just make sure that you have the right permissions to do that, and you're

sharing that with people who should have it, uh and not with people who shouldn't.

So, the notebook is a really good shortcut to just grabbing that source material together and having the context

all in one place so that you can just do that quick chat and quick summary with it. But if you're looking for something

a little more sophisticated where you really want to customize the way that it behaves and customize the types of

output and to get more detailed with it, then we can build agents. So, agents sit over here and I have got a couple in

there that I've built in this demo system before in other videos, but yours probably won't have anything other than

researcher, analyst, we'll come back to those in a moment, and this new agent button here.

Now, the concept of building an agent, even though we say build an agent, is that you are refining the Microsoft 365

Copilot chat experience to be a subject matter expert, to be something very specific. So, when I was in my main chat

experience up here and I was asking it about the Pets at Work Day, I was in there having to say, "This document,

this context, these things that are going on." With an agent, you can actually say, "You are the subject

matter expert on the Pets at Work Day. You are going to answer questions about that and this is the style that I want

you to answer in and how I want you to do that." So, even though this is called building an agent, I want you to think

about this actually as a narrowing down of the main Microsoft 365 Copilot chat experience. It's a little bit

counterintuitive, but when you are in this chat experience here, remember this is grounded on connected to everything

you've got at work, all of your emails, Teams chats, documents, the whole thing, and so you have to write really good

prompts to kind of specify what it is you're doing. The idea of the agent is that this is something that is

repeatable where you might want even more detail in that prompt and more instructions on how you want it to

behave and specify those data sources and you want to be able to kind of save that like a little job description and

make this a subject matter expert. So, I'm going to take you through this concept. When it first comes up, it

might pop up in this configure tab. I want you to go across here into the describe tab because this is an easier

place to start and I'll show you why. So, at the moment in the configure tab, this is all blank. Blank agent, blank

description, blank instructions. We're going to come back to this in a moment. So, what we're going to do here is

create an agent that can answer questions from employees about our pets at work day. You'll see I've given it a

fair bit of context here about the kinds of things I want it to do and the kinds of information I want it to be able to

answer from. And then you'll see what it does with all of that information. This is similar to the prompt that we made

earlier. So, the idea here is the journey sort of from prompt to agent and the thought process you can go through

is if you've got a prompt that starts to become quite detailed with all of that context, source, and expectations, and

you're finding that this is something you might want to reuse over and over again, it's not a scheduled prompt that

you just want to run on a regular basis. This is actually something where you now want to give it a lot of context. Maybe

you want to come back and refine it, all of those kinds of things, then an agent is one way that you can do that. So,

here we go. We now have a name, pets at work helper. It's got a purpose and we have got a bunch of little quick buttons

that we can work with. Let me show you if we go back into the configure tab here what's going on. All of this has

just been populated while while I was talking to you. So, we've got pets at work helper. Their description is like

its purpose. And now we've got these instructions. So, this is now briefing behind the scenes all of this

information about how we want it to behave. So, you'll see there's a lot more depth here than what we did in the

notebook. There's a lot more depth here than what you can do in a single prompt and it's saved and it's something that

we could iterate and improve on. So, now we've got sort of some general guidelines, focus areas, how to help,

examples of things, error handling, don't talk about things outside of that because remember this is a subject

matter expert. And then some examples. Now, you can directly edit these or you can directly

write your own instructions if you know what you're doing after you've done a few of them, but to be honest, that

description and then the agent just writes the instructions for you is pretty good way to go.

Now, with knowledge, this is really important because when you are in your main chat experience, the starting point

is that it has access to everything, your whole Microsoft 365 world of work, your work IQ.

The agent is a subject matter expert. When you create your subject matter expert, you're essentially starting with

a blank box, an empty bucket, and you opt in which data sources you want it to use.

So, if you don't add any knowledge sources here, it won't be able to access any of those documents even though the

main Copilot chat can. Remember this is a The building of the agent is the narrowing down of the experience. So, I

want to put in the things that I want. And that gives me that narrow down experience. If I don't put anything in,

it won't have access to anything. So, what I'm going to do here is bring in that whole folder of documents this time

from that Pets at Work Day that we've got sitting in SharePoint. So, if I come in here and say click to add files,

meeting chats, emails, websites, we've got all of these things here. If I click on SharePoint, then what happens is it

allows me to come in and choose a SharePoint site, but this will give me the whole site, which is not what I

want. I just want that one folder. If I click my SharePoint files, folders, whatever, that basically gives it access

to everything that I've got access to in SharePoint. So, we're kind of back to where we were with the main Copilot chat

experience. I don't want that. I want just just that one folder of documents. So, don't do that unless that's what you

mean. The way to get to that is to come over here and do this browse in the attached

cloud files. And I know that this is sitting in my communication site. There's my Pets at Work Day, and I can

just check that and select that. So, now what I've done is I've added that whole folder, and you can see that it's that

folder. So, just be careful. Don't add all of SharePoint if you don't mean to do that. Chances are with an agent, you

just want it to work with a subset of things. I can also say that I want it to add

files, meetings, chats from Teams. So, if I come in here, then I can choose to say, "Do I want it to have access to my

Teams chats and meetings?" Yes, I do, because actually we've had some of those things. Or actually no, this is going to

be something that I'm sharing more broadly with the whole organization, and so I just want those documents which are

in a communications folder on SharePoint that's public. So, that's the design of this agent. So, think about how you're

doing this. You can build this just for yourself, or you can build it for something that you're going to share

with others. And the other one here is that you can add email in here. So, again, if you want it to be able to read

your emails, if you're creating an agent where you want it to help prioritize your inbox, or

look at any outstanding tasks, you have to explicitly opt in your email. So, I'm for this one only opting in that single

folder. I could also add a website here. So, maybe there's an external website on planning Pets at Work Days, and I could

add that in there as well if I want. Use only only use specified sources. We're going to check that because I only

want it It's got in the instructions there that its remit is that we should only, you know, work with work with

what's in there, but that toggle switch kind of locks that in. This one allows you to reference the

organizational chart. For this particular agent, that isn't particularly useful, but depending on

what you're doing, that can be. I also don't want it to go out to the web. I just want this to be based on those

documents that we've got in there. And then I've got a couple of other capabilities that I can choose to add.

So, this one create documents, charts, and code. If you're creating an agent where you want the agent to be able to

output a document, then you need to toggle that on. I don't want it to do that. This is a simple kind of

information retrieval agent. But actually, it is kind of a fun topic and I do want to allow people to create

images with this one. So, I'm going to toggle that one on. Now, our suggested prompts are these

buttons that sit here and it has picked those up based on what I've done for my what I put in the description. So, these

are nice helpful things that the user can start with. Tell me about the pets at Workday, what fun activities, how do

I register, what are the rules and responsibilities. So, you can see I kind of primed that in the way I described it

to put it in there. You can have between zero and six of these. So, this is quite a simple interface. You can just choose

the little trash can, rubbish bin icon there to get rid of something you don't want. Click here to add a new one and

you can edit these. So, this is the title, what is pets at Workday, event schedule. So, let's say we wanted to say

um pet, you know, pet schedule or something like that. You'll see as I change that

that it immediately edits that over here and then back again just so that you can see what that does and then you can

change the the message. This is looking good. The last thing we might want to do here is give it a cuter

icon. So, you can come in here and choose edit the icon. This is my favorite thing here is that you can

generate it from the agent details. So, because I've got something here that's got kind of a lot of cute fun pets at

Workday type stuff in there, this should generate something. But if you've got your own image or your own icon, uh you

can choose to browse or upload that from somewhere else. We'll just pause the video while it generates the image and

be right back. All right, and there we have it. That's quite a cute little paws at Workday type thing. I can just click

apply and there it is. It's quite cute. So, once we're ready to give this a test run, you can actually test it here to

make sure it's working okay, but we're going to skip that step and just click create. So, when it's first created,

it's private just for me, but we will come back to how we can share it cuz remember this is for everyone to learn

about this. So, I'm going to go into my agent. So, this now pops up on the side. If I want to keep that there, just hover

over it and click on that little pin icon. And so, now you can see I've got these starter prompts, or I can come in

here and put any other prompt in here. But, let's have a look at the event schedule and just see what happens. So,

this should use the instructions that I gave it. It should use those documents that I gave it and come back with an

answer that is hopefully full of fairly terrible pet-related puns. Here we go. Here's the tail-wagging lineup. It

references the documents. It is doing all of the things that I asked it to do. So, obviously you would test it a little

bit more. What you can do here is continue to refine it. So, typically the journey with agents is that you'll come

in here and edit. After you've used it a bit, you'll go, "Ah, it's not quite doing the right thing. It's not quite

answering the right way. I want it to respond in a different way. I want it to emphasize this or change that." And you

can come in here and edit the instructions, or come back into describe and say, "Change it so that it doesn't

use the the puns." If that's annoying for you or whatever it is you want to do.

So, once you're happy with that, what you can do is click on this button here to say share, and then you can choose

who you want to share this with. So, for this one, we are planning to send this out to everyone in the organization.

Now, this obviously needs permission, governance, whatever in your organization. So, need to talk to your

IT people about that cuz they can have global settings here. Firstly, on whether you can use agents at all and

whether you can share them. So, if you don't have this option, go talk to your people. Uh but then this one, if you are

sharing with other people, specific users, or anyone in your organization, you will need to make sure that it's

okay for them to have permissions for the documents and things that are being shared. So, this example, I'm taking it

off a public SharePoint site, so that's a good use case for this kind of thing. And then we basically put that link out

into the world, or you can have your organization publish it so that it sits under this all agents here, and it comes

under the the place here where it's easy for people to discover. Now, speaking of all agents, there are a

bunch of agents that come out of the box with Microsoft. There are some things in here that are worth exploring. There are

things that will, if we expand the see more, some of these here like prompt coach, idea coach, writing coach are

very good at just helping you to kind of brainstorm and give you, again, it's a subject matter expert that's designed to

give you particular feedback on those things. Anything that is called frontier, and

there are quite a lot of them here. I'll have some more videos coming. Subscribe to the channel if you want to know more

about these frontier agents. So, these are um agents, if your tenant at work has opted in, you the admin's opted into

the frontier program, basically means you're getting early access to some of the things that are coming but not sort

of well-established yet. So, you're sort of like a little bit ahead of the curve with working on those, and gradually

they move from frontier into into the mainstream of the of the product. So, the things that are in there now, even

without frontier, are these two agents, researcher and analyst. These are both very high-value things, and I wanted to

take a minute to help you understand what these are about as well, because this is a very cool part as well of the

value of the license. So, both of these agents use a premium capability. Together, they're capped, at the time of

recording this is true, 25 uses per user per month. So, this is not something to use, I mean, we have about 20 working

days in a month, so you can kind of use them every day, but don't just like use this as a default. So, if you're using

it more than 25 times a month, you're probably overusing it. You don't need to use this for everything. So, the

researcher agent is something that does deep research on a topic. Imagine you need a briefing paper on a topic, or you

want to go out and sort of gather a lot of information, or understand something about what's going on. So, this is where

you, instead of giving it a prompt to say, "Here's my goal, context, source, expectations." You're actually giving it

a problem to solve. So, a prompt here might look something more like this. I'm organizing a Pets at Work Day, I'm

giving it that context of I'm in Australia. I'm asking it to put together a paper. Now, this is going to be a

little bit weird because I've got all these other documents that are already in there. So, when you run the research

agent, let's just keep it going, so you can actually kind of give it more information here. What type of

organization, the type and number of pets, I'll just go for it. I'm just going to say short, and use your best

judgment, and start the paper. Okay, so you can refine it. But, the thing that this is doing, and you'll see this in a

moment, instead of like your normal chat where you go prompt, response. Sometimes in the prompt and response, it will do a

little bit of deeper thinking, that's an option that's in there. But, this is actually doing chain-of-thought

reasoning. So, it goes, "I'm going to begin the research now." What you want to do if you're interested in this is

expand it. Usually with this, this is one to go away, have a cup of tea, or whatever whatever

drink you like, and come back, cuz it usually takes a few minutes. So, you'll see what it does. It goes, "I'm starting

to pull together the sources, compiling these things." So, I can see straight away ABC News is a the local broadcaster

here. It's going to go through like business.gov.au. So, I can see straight away that it's

picking up on kind of authoritative sources for my context, and it will pull together all of that information. Now,

I'm just going to stop this because it will go on and on, and if I go into the new conversation here, and have a look

at one that I've done earlier, then you'll see the response because it comes back with something here that's got all

of this type of information in it. So, you can see that it's gone away and done a whole bunch of research. It's got the

citations in there as well. So, there's a lot of information that's in here. And then, I can take this, if I go all the

way down to the bottom here, and choose to say open it in a Word document. And then, you will remember when we were

looking in our notebook earlier that the Pets at Work notebook actually had that document in it. The other thing you can

do though is use that researcher agent from inside the notebook. So, for the context of what I was doing here, if I

was pulling together all of this stuff and I wanted to use that researcher capability, I could actually do it in

here and then it would be part of the source documents inside the notebook. So, that's actually another little nice

crossover of those things. The analyst agent is very good at kind of analyzing data and helping you to understand

things that are going on. So, I've got a document here that I have downloaded um from a a place called Animal Medicines

here in Australia. This, in addition to having a bunch of cute images, has a lot of kind of statistics in it about

various pet ownership uh things. So, we'll give this a go because you'll see that's not actually a spreadsheet. So,

prompt for the analyst agent here would be to say analyze the data in this, help me identify any trends, help me look for

any outliers. And this one I've just downloaded that PDF. I haven't actually saved it into my files. So, I'm just

going to choose to upload that from my desktop and put it in there. And then, we will give it a moment to upload that

into a safe place and off we go. You can bring multiple things in here. So, I could also bring in the document of the

proposal and say how do these statistics align with our proposal? You could bring spreadsheets in here as well. The idea

of the analyst is where you want to bring something in where you've got multiple sources or perhaps something

that's a little bit more uh it

and quantitative together. If you want to just analyze a single spreadsheet, stay tuned because we are going to have

a look at working with it in Excel in the context of a single spreadsheet in a moment. So, you'll see what this one

does. We're going kind of going through and having a look. You'll see all the way along here we've got this coding and

executing. So, what it's doing in the background, you don't have to understand this. It's writing Python code so that

it can actually have a look at um what's going on and crunch that data in a way that my human brain certainly can't do

from looking at a PDF. So again, this one goes on for a while. I'm just going to pause the video and we'll come back

when it's finished. And it's just come back with the results. Here's a clear accessible

summary. So, here we go. What people are most likely to bring, dogs, cats. I wonder if we should have fish and birds

uh and a bunch of other things. So then, support you might want to provide. So, you'll see it's interpreted the data and

given me that kind of analysis and I bet you didn't know uh All right. There we go. Quite a lot

Quite a lot of interesting facts in there. So, that's the analyst agent. Now, Copilot doesn't just exist in this

chat experience. This is the main place where you can work and the reason to work here is if you are doing lots of

different things. So, you are pulling together information from different sources, you're jumping around to

different things. This is typically how I use it the most. But, it does follow you around wherever you go. And one of

the most common places that people find this and use this is inside Teams. So, if you are in the chat part of Teams,

you'll find you've got Copilot in here and it's exactly the same experience. You'll see you've got your work and your

web. You've got your message in the middle there. On my version, I don't have the side menu. Some people have got

the side menu down there. But, if you are using it in Teams and you're not seeing the side menu, cross over into

using it in the website or using it in the app if you want that full experience. But, I do find this is

useful in context. I do in my job work in teams a lot. So, rather than having to switch screens, the idea here is I'm

in the middle of a chat with someone and I need to check something, I can basically just chat with Copilot the

same way I'm chatting with any of my colleagues. Now, this extends into other

applications as well. And the concept of working with it in the applications is that you've got that whole experience as

a sidecar experience, but it can work in the context of the document that you're working with. So, I've got an email here

from Amy. Hi Miranda, do we have a kickoff time? Okay, I need to reply to this email. Now, there is a Copilot

built in here that'll give you summaries and if you click reply on this email thread, you'll get this option to draft

with Copilot. To be honest, I don't like this experience as much as I do over here. So, what we can do is

open this sidecar panel and now you have got all of the same stuff that we have

before. So, if I wanted to search for things in those documents, I can ask the question and put things in here. So, I

can say help me reply to this email. Find the kickoff time for our pets at work day. And again, I could reference

that document if I wanted to. So, I'm just in the context of Outlook here. I haven't had to look. I've said help me

reply to this email. It kind of knows what's there, but it's brought this in and then we've got the hi Amy, it knows

who it's from, so we've got that context in there. Remembering that this doesn't send the

email for me. Copilot not autopilot. I'm in control of what's going on here, but I can copy that response and put it in

there. Now, just be careful always when you're copying and pasting AI that you get rid of the top and tail. There's

nothing more of a giveaway that you've used AI than if it's going, "Certainly, I can help you with that. Would Would me

to ask something else?" So, just make sure that that's uh that's working? I'm not going to worry too much about the

the formatting there, but I've been able to send that. I haven't had to go away, open the Word document, find it. I can

just bring that Copilot experience directly in there. So, if I come across into Word, and I've got my Pets at Work

Day here, this is a document that I'm working on. I've got Copilot in the side panel here, and again, this same

experience. What you can do is expand this so that you sort of got a nice side-by-side experience. You've also got

the navigation panel, so if I open that, all of my chat history, as well as the agents that I was using, are sitting in

there. You can close that down again, and this allows you to do this side-by-side chat as you're creating

with the document. You've also got in here draft with Copilot, so let's close this one down for a second. This is

quite cool. So, if I've got, for instance, logic and preparation, logistics and preparation, I could

highlight this. The Copilot drafting experience within Word allows you to do various things. So, what I could do is

say turn this into a table. So, you can take sort of unstructured blocks of text and turn it into a table.

So, there we go. Now, it's given me kind of a tip and details based on what was there. You can also kind of just ask it

to generate more content for you. So, we could put a thing in here saying, "Tips for

small pets like guinea pigs." And then just ask this to Oh, we always have to click keep it. I should have

done that before. So, then what we can do here is just come in and say, "Keep writing this." And it will actually just

sort of generate content based on what I've put in there. So, you can see it's starting to give me some more

suggestions. For something like this that's very generic, this is actually very good to sort of use the AI for it,

but just giving you some ideas of other things of ways that you can draft with this.

In Excel, we have got that experience down the side here as well. So, I've got my Bring My Pets to Work Day budget.

I've got my Copilot experience down the side by selecting that. And then, I can ask it to analyze the workbook, give me

interesting insights, start to actually work with the workbook to do things and change things. And if you want to see

more on that, I've got a whole tutorial on Copilot in Excel. And similarly, in PowerPoint, you can

use Copilot in PowerPoint to generate presentations. You can also use it to sort of do research on the side and ask

questions about things and get information for the presentations and add slides.

So, you'll probably spend most of your time here in this main Copilot experience. Make sure you have a play

with the notebooks and the agents and also those experiences on the side. If you've had value from this video,

firstly, thank you for coming all the way to the end. Please remember to hit that subscribe button if you're not

already subscribed and add a comment to let me know that you made it all the way. I always love seeing those ones.

I've got another video here that takes you through how to make Copilot more personalized for your experience.

Recommend checking that one out. Thanks for watching.