Microsoft Copilot: A bitesized overview for businesses

Most businesses have access to Microsoft Copilot, but many still aren't sure what it actually does or how to use it.

This beginner-friendly deep dive led by Shimron Walters will walk you through Microsoft's automation ecosystem, AI agents and the capabilities that can transform the way you work.

Topics covered in this session

• What Microsoft Copilot is, how it works and why it's quickly becoming the industry standard

• How Copilot Studio empowers you to build custom automations and AI agents tailored to your business

• Practical use cases that show how this technology can help your team save time, reduce costs and operate more efficiently

About the speaker

Shimron is the founder of BitesizedGenius, helping businesses solve problems with tools like Microsoft Copilot to save time, money and operate more efficiently.

Through tailored training, hands-on consultations and the development of solutions, he supports organisations to embed AI into everyday workflows in a way that delivers real value and supports responsible use.

Watch more expert webinars

Access a growing collection of expert-led webinars covering marketing, sales, finance, growth and more – ready whenever you are.

Transcript

Lightly edited for clarity.

Caitriona: Hello, everyone, and welcome to today's Lunch and Learn. My name is Caitriona, and I'll be your host today.

For those of you attending a Lunch and Learn for the first time, Enterprise Nation is a vibrant community platform for start-ups and small businesses.

I'm pleased to introduce Shimron Walters, founder of BitesizedGenius. In this session, Shimron will walk you through Microsoft's automation ecosystem, AI agents and the capabilities that can transform the way you work.

If you have any questions throughout the webinar, please post them in the chat, and we'll do our best to answer them at the end of the session.

Today's webinar will be recorded, and we will send a follow-up email with the recording and further resources later today.

Over to you, Shimron.

Shimron Walters: Thank you, Caitriona, and good morning or good afternoon, everyone. I hope you're all doing well.

I've got 30 minutes to go over something that I usually spend about two hours on, so I'm going to try and condense as much information as I can into what will hopefully be a very useful whistle-stop tour of the Microsoft platform, specifically looking at Microsoft Copilot and some agent capabilities as well.

Hopefully, you can all see my slides on screen. There won't be too many slides because what I'd like to do is go over some high-level information and then move into a demo so you can actually see things in action, rather than just looking at slides.

So, without further ado, we'll get started.

A bit of an introduction about myself. I'm Shimron Walters, founder of BitesizedGenius, and I help businesses with their understanding and adoption of AI, specifically within the Microsoft ecosystem.

The main reason for that is that a lot of organisations already use Microsoft products such as Office365, but they also worry about security, customer data, and how they can work with AI inside a secure environment.

Do feel free to introduce yourself in the chat. I'll keep an eye on it and see if there are any questions people want answered.

So, I'm going to get started, and the first thing I wanted to quickly distinguish is that there is a difference between Microsoft Copilot and Microsoft365 Copilot.

There are actually two versions of Microsoft Copilot, and a lot of people aren't aware of this, which can sometimes cause confusion.

When we're talking about Microsoft Copilot, it's essentially the equivalent of Microsoft's ChatGPT. You ask it a question, you get a response, and it has AI capabilities that allow you to get information from the web and from documents as well.

Then you've also got Microsoft365 Copilot. This is actually linked to your workplace data, so for example your SharePoint and your OneDrive.

It's more secure because it's grounded to the data that you already have in Microsoft when you're paying for the subscription.

The way you can tell the difference is that if you look at the logo, Microsoft365 Copilot will have "M365" with a little black box, whereas regular Microsoft Copilot just has the Microsoft Copilot logo.

Those are two key distinctions I wanted to make because these two platforms do have different capabilities.

The key difference is that Microsoft Copilot is more of a chatbot user interface. You ask questions and get responses. Microsoft365 Copilot is closer to ChatGPT in the sense that you can create custom agents and you've got a lot more capabilities there.

I'll show some of that when we get into the demo. I just want to quickly go through some information on the slides first.

So, if I bring up Microsoft Copilot here, this is the place where you can ask a question.

Shimron Walters: So this here is Microsoft Copilot. You can ask it a question, get a response, and it also uses computer vision, where you can share your screen and it can look at your screen, provide information, and answer questions based on what you're sharing.

The one I'm mostly going to cover today is Microsoft365 Copilot, which is this. This is more like the ChatGPT-style version, and we have a range of different options.

The first one we have is the difference between the Work and the Web tab.

The difference is that with the Work tab, it's grounded to the data that you have on things like SharePoint, OneDrive, OneNote, and other Microsoft information management systems, so that you can interact with those documents to get information.

With the Web tab, you can ask it a question and it will use internet search, I think through Bing, alongside information from your documents.

You can also use web search in Work as well, so there is some overlap. But the rule of thumb is that if you're using this for work documents that you've got on SharePoint or OneDrive, then use the Work tab. If you want to conduct an online search alongside using your documents, then use the Web tab.

We've also got two keys you can use in here: the @ symbol and the forward slash.

If you use the @ symbol, this allows you to interact with an agent. There are some that come preinstalled, such as Researcher, Analyst, and Microsoft365 Admin.

You can interact with those, ask questions, and use them for different tasks.

You can also use the forward slash key to search for documents that you have on your information management systems. You can interact with those documents, begin asking questions, and work through the information that might be useful for your research.

Now there are some other features that we have here. A lot of people are also unaware of this, but Microsoft365 Copilot can create various documents.

So, for example, if you wanted to create a Word document, an Excel spreadsheet, or a PowerPoint slide, you can do that.

Let's do a quick example. I'm going to ask it to draft me a poem. So it creates the poem, and then what I can do is ask it to export that into a Word document.

What it will do is draft it into a Word document, and then once I click on that link, it opens it up directly in Word.

You can do the same thing for Excel spreadsheets. If you were trying to get on top of budgeting, for example, you could ask it to create a budget spreadsheet based on your expenses and bills, and then open it in Excel.

Microsoft Copilot can also create images and other forms of media, such as video. If I go over to the Create tab, you've got a range of options here: videos, infographics, storyboards, and so on.

That said, there are also other tools you can use. Midjourney is very good for images. Google's tools are getting good for images and video as well. ElevenLabs is excellent for audio, transcription, narration, and audio-related use cases.

So even though Microsoft Copilot can do these things, the quality can vary, and there are other tools that are more specialised for certain activities like audio and video.

Then lastly, you've also got notebooks.

Notebooks are essentially a way of adding lots of information, or different documents, into one place so that you can work on them together.

You can create a notebook, add different files into it, and then work with all of that data in a place that is isolated and grounded.

This doesn't replace OneNote. You still have access to OneNote, but this is another way of grouping information together and working on it.

Now something else I wanted to quickly go over is the agent side of things. So if I go to the Agent tab on the left here, just beneath Chat, there are some agents that are already available.

There is a difference between the types of agents you get built into Microsoft365 Copilot and the ones you can make yourself.

There's also a difference between making something like a simple chatbot agent, where you give it your own documents and information, and a more autonomous agent.

So if I go to Create Agent, those of you who've used ChatGPT might be familiar with a screen like this.

You can create your own agent, give it instructions, and give it knowledge sources in the form of website URLs, documents, meetings, chats, emails, or SharePoint locations.

You can then essentially get context-aware responses to your questions based on the documents you provide.

So, for example, let's say you work for a company and you've got lots of frequently asked questions. Employees might ask about payroll details or company policy.

You could create an agent that contains information about payroll, HR, and all of those things, and then people can interact with it and get responses that are grounded to that information.

You could also create documents, charts, code, and images using these agents, although I usually disable image generation because it's not something people tend to use much in that context.

Just as a quick example, I'm going to create a simple agent. I'll call it Example Agent, and I'm going to say this is an agent that responds using bullet points.

The description is for our benefit so we can understand what the agent is doing. The instructions are the behaviour of the agent: what the task is and how it should complete it.

So I'm going to upload a document to this agent. This is a small document I made called "Format", and inside it, it says, "Respond to all questions using bullet points only."

The reason I've added that in is so that when I then ask it a question, it will use that uploaded file as part of its context.

So if I save this and then ask it to write me a poem, it should respond in bullet point format because the document I attached told it to do that.

And that's exactly what it does. It says that because the uploaded format includes guidance to respond using bullet points, it will shape the poem in bullet style while keeping it lyrical.

So you can see how it goes through the uploaded files, looks at the information, and uses that as context.

That could be for anything. Company HR policy, research documents, internal guidance, whatever you want.

A lot of people would describe this as an agent, although I'd call it more of a customised chatbot. For me, agents are a little more autonomous. They complete tasks and can run more end-to-end processes.

But that gives you a whistle-stop tour of Microsoft365 Copilot and some of the features available there.

Now I want to move on to some of the more agent-focused things, because I know that's what a lot of people are interested in.

So we've looked at Microsoft365 Copilot agents, which are those custom chatbots that you can ground in your own information.

Now we've also got Microsoft Copilot Studio agents.

Before I move on, just one last thing. If you go to your settings, then to Chat, then Manage Settings, you can access Personalisation.

Custom instructions are essentially a way of tailoring how Copilot responds to your questions. For example, you can say things like "give honest feedback", "use clear and simple language", and so on.

You can customise it however you want. You can tell it to use bullet points for summaries, avoid emojis, speak in a particular tone, whatever suits you.

You also have memories. Memory allows ChatGPT or Copilot to remember certain preferences so that it can provide better responses over time.

Personally, I tend to keep this disabled, because I've found that sometimes AI remembers things I don't want it to remember.

For example, I might ask once for something in bullet points, and then it decides I always want things in bullet points, which isn't true.

So I usually prefer to prompt what I want each time and work with it more like an assistant, rather than letting it learn too much about me.

Now, coming back to agents, we've looked at Microsoft365 Copilot agents, and we've got Microsoft Copilot Studio agents as well.

What I'm going to bring up now is Copilot Studio. This is how you create workflows on your data, such as email drafting, data manipulation, website chatbots, and more.

For me, this is really where the power of Microsoft's agent capabilities comes from, because the number of things you can do in here is quite overwhelming.

There's a lot of talk at the moment about everyone becoming a coder and everybody building their own products.

The reality is that there is a learning curve to all of this. You have to invest time and energy into understanding how these tools work.

Different tools have different levels of difficulty and ease of use, but all of them have a learning curve.

It's a bit like when Microsoft Excel first came out. We all had to learn how spreadsheets and formulas worked. It feels second nature to us now, but at one point there was a learning curve. That is still the case today with AI tools.

So I have an agent here that I made, and this is a simple agent that can do two things.

First, it can answer frequently asked questions. So if you put this on your website, for example, on a clothing store, it can answer customer questions based on an FAQ document attached to it.

Second, it can get updates on customer orders.

The way it works is that customer order statuses are uploaded to a database in Microsoft Power Platform, and then the agent can extract information from there based on what the customer provides.

So if I click the button that says I want an update on my order, I can enter a customer reference number and a date of birth, hit submit, and then it will give me the order status.

For example: "Order delayed, awaiting restock, expected within 48 hours." Then it takes you back to the menu.

You can also configure this to speak to a human, and Microsoft Teams also supports that, so you can engage your customer support team that way.

You can also ask FAQ-style questions such as "What are your store locations?" and it will go through and say Oxford Street, Paddington, Brixton, and so on. Or you can ask, "What is your return policy?" and it gives you the relevant answer.

So instead of calling customer support, the customer can interact with this. But they also have the option to speak to a human if they want to.

That's really important, because I'm sure if anyone has ever dealt with HMRC, the worst scenario is when you've got a question the AI can't answer but you also can't get through to a real person.

So the human fallback matters. If we look behind the scenes, you've got knowledge, tools, and a lot of advanced functionality here.

You can have child agents, topics, and the topics are essentially the different types of questions that can be asked, the responses you get, and how the interaction is structured between the user and the agent.

I also want to go to Flows because this is actually one of the really key things.

Flows are where you can customise sets of functionality. For example, extracting information from Dataverse, working with variables, if statements and loops.

There are some basics of programming involved in this. It can get quite deep. But in terms of creating end-to-end processes, this is really where some of the power sits.

Now, in the chat, I can see a question about what you actually have to do to build an agent, and whether you can ask ChatGPT to write those instructions.

You do have the ability, when you're creating a blank agent, to use an AI-assisted feature that allows you to start by describing what you want the agent to do.

You can fill that in, and it will give you a starting point that you can then customise.

But it still does require an understanding of how some of those features work. You can tailor things to your liking, but there is still a bit of a learning curve.

Because we've only got a few minutes left, I want to leave some time for questions.

So I'll quickly move on here.

Please do visit the QR code. Feel free to ask questions. The QR code will take you to my website. Hopefully you can also get through to an eBook at some point as well, which may have some useful information.

I'm very happy to answer any questions.

Caitriona: Thank you so much. Let's try to do as many questions as we can.

We have one question here from Bethlehem. To use Copilot, does it need any registration?

Shimron: Yes, there is a registration process. If you type Copilot Studio into Google and go to the website, you've essentially got a few packages there.

You've got one that you pay for, which is Microsoft365 Copilot, and that allows you to create agents that can be used within products like Word, Excel, PowerPoint and so on. You've also got a pay-as-you-go option.

Microsoft are not brilliant at describing how things cost and how much it actually costs to set up, but my understanding is that you can enable the pay-as-you-go option and then start creating agents.

Those pay-as-you-go agents can't usually be used directly in something like Word or SharePoint. They're more automation-based and tend to run on Azure, which is Microsoft's cloud platform.

I'm happy to have a chat about it, because it is something you do need to look into before setting it up. But once it's set up, it's fine.

Caitriona: Thank you. We do have a lot of specific questions, so it may be best to include the QR code and encourage people to book a chat with Shimron.

But let's look at this one.

"I have a bookkeeping question. Can I upload my expense receipts to Copilot and get it to capture the receipt info and extract it into a spreadsheet?"

Shimron: Yes, that is entirely possible. What you can do is use image vision. In AI, you've got the ability to extract text from images.

That would be useful for uploading a receipt and getting it to extract the information.

The next question then becomes, how do you get it from your phone into SharePoint or wherever you want it to go?

It might be a case that you use your phone to take an image of the receipt, save it somewhere like SharePoint, and then have an agent scan the folder of receipts, extract the information, and put it into something like an Excel spreadsheet or a Power Platform table.

So yes, it is entirely possible. The more complex part is how you go about building that solution. That's where you'd probably want to speak to someone like me.

Caitriona: Thank you. We're just coming to the end of the session now, but we do have time for one final question.

What's one overlooked feature of Copilot that more people should be using?

Shimron: Definitely Flows.

Most people I speak to who are curious about Copilot are not experts in it. Most of the experts are probably at Microsoft because it's still relatively new technology.

But in terms of the features that are overlooked, definitely Flows.

I actually have another agent here that lets me research local businesses using data from Google Maps.

So I can type in where I live and the type of business I want, and it can return information about local businesses such as law firms or accounting firms, relative to my postcode.

It can do automated research, such as what the company does and who the key people are. That all gets created into a Microsoft Word document automatically. That functionality is powered through Flows.

So being able to use Flows, understanding how to use variables, if statements, how to work with tools and APIs – this is where a lot of the power comes from.

And I think in future, as we continue developing in a more digitally enabled world, we are going to need to keep upskilling ourselves on becoming more digitally fluent. It's becoming an increasingly important part of our work.

So yes, definitely looking at Flows, tools and APIs so that you can extend functionality beyond just the Microsoft ecosystem.

Caitriona: Thank you so much.

Please do connect with Shimron. I put some links into the chat, and there was also the QR code. We can include a link to that in the follow-up email as well.

Thank you so much for joining us on today's webinar. Thank you, Shimron, for your presentation.

Explore our webinar library

Unlock more on-demand sessions designed to help you sharpen your skills, grow your business and stay one step ahead. Find more Lunch and Learn webinars