View Recording: Boosting Productivity with Microsoft Copilot

Corey Milliman 0:04 All right. Good morning, everybody. Looks like we have a good group so far. And as more people filter in, we’re going to go ahead and get started here. So today we’re going to go over copilot. We have some new things that have been released, of course, in July, August, September. And then we are also going to demo some of those new capabilities and build some agents in Copilot today. So if you have any questions throughout the presentation, just make sure drop those in chat. I’ll try to monitor those. Other folks at Concurrency will help me with those as well. We’re going to go ahead and get started. So some new capabilities that came out of Copilot. So we have the Copilot control system that’s been rolled out. We have broader SharePoint agent reach so that we have those agents available across the organization without going back and forth between Copilot Studio, Copilot and different websites. Now we have image reasoning and editing, custom dictionaries were released, and of course we have our new model, ChatGPT 5. So let’s just get right into it. So first of all, we have this new Copilot control system and agents included in the admin center. Internally developed agents that were deployed Microsoft 365 services user create agents within copilot studio in copilot chat and our 3rd party agents. So our SharePoint agents can be discovered across the organization. We can manage those. We can make sure they are governed within the agent section. In the Microsoft 365 Admin center. So this allows us to also do different things like review the metrics for these to understand how they’re being used, and we can also disable these, understand how they’re being used. Transfer ownership even and do some other things there as well. And in those advanced metrics that we see, we’re actually also able to review those metrics and look at the usage reports for these. So we’re going to be able to see our adoption trends of these user defined agents. We’re going to give insights that enable admins to make more data. Excuse me, data driven decisions to optimize engagement, identify user enablement opportunities and ensure that copilot search capabilities are delivering maximum value across the organization. So along with that, that feature rolled out in August and to support governance and compliance, Microsoft the Admin Center now. Also includes a request and approval workflow for frontier as well as agents of Microsoft 365, so the frontier program for copilot offers administrators really the ability to get hands on with the latest model innovation and provide feedback. Before experiences are made generally available, so it also gives our administrators a centralized, audited process to review and approve these agent request prior to our users being activated. So this way now finally with our user created agents. We have a controlled, transparent approach that balances innovation of our user community and our organizational policy requirements. So the approval flow for Frontier and Microsoft 365 agents will be available under the admin panel under Admin, Center, copilot. Agents. And then you go over to requested agents so that feature also rolled out in August. And then it looks like my slide presentation is stuck. Let’s go back here. So we also have our new GPT interface and so and our, I’m sorry, we have ChatGPT 5 and our new interface, so. A new model and new model reference options for copilot chat. Now we can take that users prompt to understand it and use GPT 5 ‘s real time router to choose the best model to reason over that prompt and crafter response. So the ChatGPT 5. We have 2 different modes. We have a deep thinking mode. And we have a kind of a quick answer mode. So by using this auto mode copilot now can understand the intent of the user ‘s prompt to understand how deep it needs to go. So that’s really really interesting and the users can actually click try GBT 5. For a session. So for simpler task copilot will use GPT ‘s high throughput model and when it identifies a complex task, copilot will automatically use GPT 5 ‘s deeper reasoning model and that also came out in August. So then we’ve also have a new tools menu that’s rolled out and this new tool tools menu in chat makes it easier to access AI capabilities right where users need them. So it’s actually located in the user chat down here. And you can see we have this tools notification there and this gives us a way to discover and launch features like designer pages, pinned agents all in one spot. So historically, we’d have to go over to our sidebar and do different things and pin those. So this helps us service some of its surface rather. Of the advanced capabilities directly in our interface that also came out in August. And another feature that came out in August was when referencing an email in copilot chat, users can now receive responses that incorporate insights from content attached to the email. So before we were looking at the email message, but now when I ask about an email message, not only does it understand the content of the email, but all the attachments. That are associated with that, so it could be a Word document, Excel spreadsheet, PowerPoint, PDF or even. It even works with structured data like Jason and XML. So copilot is going to be able to reason over email attachments alongside the body of the email to deliver a better response. So again, this feature is available when using copilot chat in Outlook or in Microsoft 365 Copilot app for users with a copilot license, and that was another new feature for August, and now we have some PowerPoint acceleration. So. When referencing this PowerPoint document, when we’re actually going through and want to create something quickly on the fly, we can actually have a conversation and right in chat it’s going to download or prepare a presentation that we can go ahead and download. So if we want to prepare just a few decks for a quick team meeting or internal update, I can give copilot a quick prompt directly in copilot chat instead of opening up PowerPoint and doing different things like that, I can have a real quick PowerPoint presentation. Generated on the fly. So we also have enhanced image reasoning and editing and these are built features that came out in August as well. So now natural language can be used in copilot, will generate new images or make more precise edits to existing ones. So we can actually change images now instead of just generating. So you can upload a photo. Talk to copilot and ask questions about it. Extract details and you can actually start iterating against multiple changes here so you can see here that we use copilot to change the colour of the graphic in the background and make some other minor changes to these images. And you can actually start keeping track of everything that’s been added when creating or editing your visuals. So when new content is added, an attachment snippet will be displayed and this makes it easier to manage and reference your visuals and the history of how you’re creating those. The other big one is custom dictionaries, so this came out in July actually, so IT administrators can upload custom dictionaries for their tenant that supports English, Spanish, Japanese, French, German, Portuguese, Italian and Chinese. So this is accessible through the copilot settings page in your admin center, and by integrating organizational specific terminology. Custom dictionaries help your AI models across the organization produce better content. So if there are terms that you usually use inside your organization that AI struggles with by having this custom dictionary, now this enhancement benefits copilot and intelligent meeting recap, ensuring that your unique business language is recognized and reflected across the copilot stack. So it helps reduce misunderstandings. It helps make your meeting summaries and teams and directions better with copilot. And it actually also is a really cool feature. Again to help understand. Business specific or organizational specific lingo that you use and and then you don’t get these hallucinations from copilot when it doesn’t understand what you’re talking about. So that rolled out in July as well. So that covers a lot of some of the new features. There are many, many more, but those are the big ones. So again, today we’re going to be focusing on copilot agents, and these are agents that are created and available to users that are done either through SharePoint or through a copilot license. So we’re not talking about copilot studio agents today. We will be covering. I will be covering those and concurrency is covering those in future sessions, but I have a big one on those coming up in October. So the Co pilot agents are really divine defined to help take on these mundane and repetitive tasks and with all of these new enhancements and improvements. These agents that individual users can create are becoming more and more powerful with each new release, so these agents can really be personalized to the domain, the knowledge experts and the domain owners across the organization. And again, as I mentioned, now we still have our centralized governance and controls. Finally over these agents. So we can feel a lot better about rolling these out to the organization. So some of the agents that are available out of the box that you’ll see our researcher of course analyst and all the other ones here that are listed in this deck that I have today, it’s kind of a take away deck. So we’re going to go through a few slides. And then the rest of the webinar is going to focus on real usage and then you will have this all as a reference. So there are a few steps to go through and building agents in copilot chat. These are documented here as a takeaway. So you have a step by step, and we’re going to go through that today. And then we also have the ability to create agents directly in SharePoint and we’re going to go through those today as well. So today we’re going to demo creating an HR agent. We’re going to create a new product advisor and this product advisor is tailored towards helping us develop a mobile app and then we are going to create a custom return analyzer. So I’m going to go ahead and stop the slideshow. And then we are going to go ahead and. Jump in here. So I’m going to just move things. All some difference here. And first, I’m going to see, I see we had a question. So I want to review that before we jump into the agents here. I see. Ohh the agent. The AI assistant was the only notification that we have so far. All right, so here I’m bringing this over to this other monitor. So it’s a little bit easier for everybody to see today. And then because teams sometimes does this when I’m sharing multiple screens and my camera, I’m going to turn off my camera for a few minutes so teams can be a little bit more responsive for me. So all right, so the first agent we are going to create here is our dynamic ask HRR agent. And what I’ve done is I’ve created a site that has quite a few documents already. That we are going to use for our SharePoint demo today or our agent demo. So in this HR document library I have about 38 different documents that are out here that cover all these different policies for an organization. So. We have holiday schedules, insurance information, you know, a lot of the normal things we would generally see and for today’s demo, I’m pointing our agent at a document library, but we can also point this agent at an internal SharePoint site. You know, our public facing internal HR site. Or really any site where you want to create the ability for self service answers. So if there’s a site you work with a lot and people ask you a lot of questions and you keep saying go check this document, here’s a link. Here’s the answer you need. You can start surfacing these agents so people have the ability to be more self service with what they’re looking. So to get started you can see here that I have my agents over here on the right hand side of the screen and I’m going to start by creating an agent and this dialogue is going to come up and allow me to start with a new agent. Since we wait for this to catch up here. And we are going to give it some instructions on what this agent is going to be. So you’re always prompted with a describe and you’re also able to choose from one of these predefined templates if you want to start that way. So these templates are expanding. But I’m not going to use the template, I am actually going to just give it a very basic prompt. Right now I’m going to say human resources agent there actually copied this over. I do have some instructions here for it and I’m going to drop those in telling it that the what its role is. So you’re going to provide employees with quick access to policies, answer common questions, provide policy enforcement guidelines and help employees navigate benefits, insurance, payroll. And then I’ve already. So now it’s going to come back and it’s going to ask me what I want to name it, and I’m probably going to. Here we go. So it’s tells me it’s been set up. I’m going to go ahead and just jump over to configure. You can have a conversation with the agent and refine how you want to work. So I’m going to go ahead. Veridian dynamics ask Ahr. It already has my description, and it does have the instructions that I previously provided. So now I need to provide it with my knowledge source. And you can see here it’s going through and. Going to my recent documents, files, sites, chats, those are all different things that I can use for knowledge for something. So I have a link that I copied over here and this goes directly to our HR policy library and one thing I this is new is to prioritize the knowledge sources I add. For knowledge based queries, so this is important because we are prioritizing our internal knowledge over the generalized knowledge of the AI. So you can see here I do have some other capabilities that are available. We’ll use those in a different in a different agent today, but it’s also based on my knowledge, sources gone ahead and provided me with some default prompts that we can use here. So if I really don’t want, you know, one of these over here, I can actually click the. There we go. My screen just hung up for a moment. Here. It’s auto saving and once this is done we can go ahead and click create here. So I did go ahead and just delete one of those so my tiles automatically updated and you can test your agent before creating. And what is? OK, I’m gonna say how do I contact a payroll provider? And this is just a way to test how the agent is working before creating and sharing out with users. And so when you do run the test it is a little slower than after you click create and it’s telling me who our payroll provider is. Here are the contact deals details ADP. Here’s the phone. Number here’s the website and it says let me know if you need help with the specific issue and then it also is giving me a link over to a document that has all of our payroll provider contact information. I’m gonna go ahead and click create. And one thing that’s important to note when we’re creating these agents is that if I share these with users, the users need to have access to the content behind it. If you want a user to be able to use it. So if I am sharing. If I create an agent that maybe is based on some information, I only want for one department and somebody inadvertently shares this agent with somebody out of that department. The agent will not work for that other individual because they do not have access to the data that was supplied to create the agent. So it’s respecting. SharePoint permissions on that SharePoint site and so that way if somebody shares an agent they can’t use it because they don’t have access to the content on the back end. So again we have these same governance and security measures that have follow our agents across the organization. So here you can see that now my agent has been published. I’m going to go ahead and pin this one here and then from here I can start. Of course. Anybody that has access to this if I’ve shared this out can start using this and ask questions based on the 30, some 40, some different policy documents across my SharePoint site. If this was pointed at my entire human resources site, it would also be able to use site libraries, lists and other content on the SharePoint site and all of the documents on that SharePoint site to generate answers. So from here I can start asking questions of course about. What I can just say I? Just got married and need to switch to a family insurance plan. What are our offerings? Or. Family coverage. So we can start asking different questions about our policies, procedures and this just shows that any of the information that is out there now it is understanding our policies and procedures, understanding my intent. It’s actually saying hey, thanks for your message and Congrats on getting married. So and it does understand that marriage is a qualifying event and you’re now eligible to switch. To a family insurance plan. And here’s an an overview of the organizational benefits that are available. And now it’s actually giving me information. How can I compare hdppl and hdhp for my situation so it is more powerful than it was even 90 days ago as far as how it’s interacting with our content and it’s doing a much better job of giving me information creating tables for me creating comparisons. And giving me the information I need in some in a way that is easier to consume than it was just a few months ago. So it’s giving me even some key considerations showing me my lower deductibles, comparing the plans for me, and it’s asking if I want to estimate my total costs. And I’m going to say, who do I contact to switch my coverage? So it’s not only understanding policies and procedures and now it understands that in our organization we have a benefits manager and here’s his email address. Here’s the telephone number and he’ll be able to guide you through the process. So I can go ahead and start an email here. I could even help it draft an email for me and just say draft an email to John Smith for me asking about switching benefits. So now it’s going to go ahead and do that for us here. So an example of a multi step and then I’m going to take this and I’m just going to it, pulled my name, pulled my contact information, says I’m out of the Chicago office and and allows me to copy this and drop this into my email here. So another window I can drop this into my my email and send that off over there. Super simple. So it’s just one example of the HR kind of interactions that we have and that is based on all of the content again that we have across our. Or different documents that are out there. I do see a question here and I want to pause and take a look at that. Can you create the HR agent to only use the internal knowledge? Absolutely. So we are tying it over to our knowledge sources and that’s why in our we’re going to go back over to create an agent and. And this is also how we go back and modify an agent so you can see here I went create an agent. I’m going to go over to my agents. It’s going to show me all the agents I’ve created. I’m going to go ahead and click the pencil. And we can do this 2 ways. Prioritize the knowledge sources you added for Agent knowledge based queries and then we can also add additional prompting that says these do not make up information and when responding to. Employee requests only use the information in the knowledge sources associated. Don’t judge my typing skills today associated with. This agent so I haven’t changed the spelling here. And just by modifying these instructions a bit, we have up to 8000 characters we can use, so you can also use your copilot license to help you write a better prompt to help you do the. One way is now I copied out the instructions I have. I’m going to go over to copilot now. I’m gonna say I just created an HR agent. I need a better set of instructions. Here is what I have so far. Please write an instruction set up to. 8000 characters to guide my new agent. I want to make sure it really uses internal documents, doesn’t generate answers to please users. And all answers are grounded. And I’ve misspelled that same thing again, but here. So I gave it my base prompts that I had created already. Now this is a really good way to use copilot to help you create agents, because now I gave it a little bit of information about my intent. And now it’s going to go through. Look at my what I gave it as a prompt. It’s going to generate a new set of instructions for my agent and I am. This is very short and we do have our HR agent instruction set that it created and I can download this file. And unfortunately, with my monitor, half of it is where I can’t grab it. So we’re just going to do the copy paste, see if I can grab that window. So here is the what it gave me so it did give me a full downloadable instruction set and so here I’m just going to copy and paste this. I’m going to go back into my agent and create agent. Going to go back and go to my agents. And now I’m going to update my instructions gave me or just just over 3000 characters. I’m going to update it and I see that there is also a another question here, is there a limitation on file size and the knowledge source? Especially for Excel files, I have an issue with the agent getting information after X rows from Excel. So if you have a lot of Excel documents in a document library, there isn’t an explicit limit on file size, but that is where sometimes. Better to open the document in Excel and use copilot inside of Excel to chat with that document or to use GPT 5 and then you can tell it to think more deeply and make sure it accesses all rows so there’s not a hard limit. But once you get over about 1,000,000 rows, it really has trouble with that. And if you have many, many spreadsheets that you’re trying to reason over, that can get dicey as well. So if you sometimes it is a little bit better and we’re going to go through one of those, I have a couple of spreadsheets that are pretty large that we’re going to use for customer churn analysis. Just show how advanced analytics and Python mode in Excel helps us kind of move past some of those those different issues there. Because yes, you’re absolutely right. Sometimes after X rows, it doesn’t always pull back all the data that you need. So now our updated agent has been that has been updated here with more information. What is? Our leave, I see. And now it’s going to go back and get our our formalized leave policies and what those policies look like. I purposely didn’t ask about FMLA or bereavement or different things there to show how it’s going to go through and show the different types of leaves that we have available. We have paid time off. We have parental leave, we have ethanol a, we have bereavement. So it understands we have all these different types of leave policies and so I can go through and ask questions specific about those. So another question I’ve had agents I’ve tested using files nested in folders I’ve had to manually move. Not think everybody can see the the questions in chat here. So if you’ve had some issues with agents in nested folders and you’ve had them manually move files out of subfolders into a single folder to see it properly, that does happen sometimes. I’ve had issues with agents not being able to create files properly when compiling data. As it does in the normal chat menu and do the agents use GBT 5 so the agents do have the ability to use GBT 5. They haven’t surfaced that as a actual. Option when you’re creating those agents. When you are getting into very complex data sources and structures, some of these things that you’re running into means your agent might actually be a better candidate for copilot studio. Instead of trying to do it all through copilot. So when we have very complex data structures and large data sets, we are trying to read out reason over and we need to get more granular and have better rules and make sure we can adjust all of our nested folders. All of this different content be specific about which models we’re actually using for different types of content. And driving topic based discovery decisions and actions, that’s where copilot studio sometimes is better. So these agents are really good at allowing our end users to create things to help them in their daily workflows. But when we’re looking at analyzing very large complex sets of data across multiple folders in large SharePoint libraries, some of those use cases are better suited for copilot studio agent instead of copilot like copilot Studio Light. With these copilot agents that we’re creating in this in this scenario. So if you’re hitting some of those things and you’re having to reorganize information and you don’t want to have to reorganize information and go through all that, you might be better off going copilot studio. They have pay as you go. You have a message packs per month. You can also purchase. I believe these starting is $200 your user license. Is to create. Those are free, so and then there are different things depending on your knowledge sources. If that all that information is in SharePoint and your users already have a copilot license, then we can let them use a copilot studio agent without incurring messaging fees. So one point and that jumps to another webinar, but if you do start going the copilot studio route, as long as your author and all your users have copilot licenses and your knowledge is being inside of your tenant in SharePoint. And you’re surfacing that agent only through teams or SharePoint site. You don’t have any consumption fees. Depending on the model you use. So there are different models that do have some different things that you can that they will charge messages on, but that might be a better way. To handle some of those advanced use cases. Any other questions on the HR agent or pointing an agent towards knowledge to provide self service to different people in the organization across it could be anything, could be our marketing collateral, it could be a standard language reuse, it could be. Looking at branding specifications to understand the colors that we’re supposed to be using in our communications. So any questions about how we could use these or the HR agent is just an example that a lot of us are familiar with so that we can show how we can use a copilot agent to service information from SharePoint sites or. Documents in a way that is self service chat oriented like this. If there are no other questions, I am going to continue. With our next agent that we’re going to develop today, so the next agent that we are going to create is to help us with mobile app development. So I have created a pretty large. We have a pretty large data set that shows I don’t remember the point in time but shows all of the apps that are available in the Google Play Store and it also has ratings associated with that. So we have 2 different data sets. Let’s we have a list of apps. We have their average rating, and then I have another data set that’s actually showing what those ratings actually are. So I’m going to open that up so you can see that before we start creating our agent. Right here. So it’s on the same site to make things easy for me today. But I do have a data analysis from the Google Play Store. So we have user reviews. And the reason for gathering all of these user reviews you can see we have some problems with language and we have some bad data and things here, but I do have. Quite a few rows of data here, so this is a very large data set and it’s by categorized, categorized and showing positive, negative and user sentiment. So I have my overall sentiment. I have a sentiment polarity. On a sentiment subjectivity and these are all the metrics that are available in the Google Play Store. So I can start understanding overall overall consumer sentiment about different apps that are out there. And the reason that I’m pulling this into the data set. Is that for this agent? I want to develop a mobile app. I want to understand the Google Play categories that are out there. This is for Android users. In this scenario I want to create an app that is actually maybe a unserved section. Or underserved. That is not a huge lift. I don’t want to create a Facebook. I don’t want to create an Instagram. I don’t want to create a maps application, but I want to look at their user reviews to understand what other developers are doing. Well, what they’re not doing well to make sure I make an application. That may be out of the gate is going to be competitive in a very busy landscape. So we’re going to go ahead and go ahead and create this agent. And I have some of my language already defined in some other windows here, so we’re just going to call this mobile app development. And I have my full prompt for this over in in the S code. And so I’m going to let you know what its mission is. It’s an Android app market analyst, and this is for identifying high value, low offered mobile app on opportunities. So I’ve given it a very clear set of instructions. I went through what is I need to tell to me to the agent what’s high value. For me, what’s low effort? Because a 2 person dev shop versus 1000 person dev shop that’s very different based on who’s using it. How is it going to analyze the data? So look at category, performance, sentiment, mining, competitive landscape. I’m going to add some additional websites to actually look at the competitive and landscape I’m telling it about. My target revenue models low offered app types to prioritize high effort app types to avoid how I want my analysis done. And specific analysis instructions. So when analyzing the Google Play Store cross reference downloads with revenue estimates calculate revenue per download, you know. So I have a very, very strong prompt here to get us started. So we’re going to go ahead and we’re going to bring this in and we’re just going to say. And jump over and I’m going to give this my mobile. You know typed. Going to leave the description I came up with. We’re going to go ahead. My prompt is just over 6000 characters. I’m going to go ahead and start getting it knowledge now so my other window. I do have some links so that these are available immediately. So I do have my SharePoint site that has 2 Excel documents. You could actually have more Excel documents in this scenario. I’m also going to add some competitive information, so I’m going to go ahead and add some external sites so I understand what’s trending in the Google development world. And Android development. And I’m going to look at pulling in what’s going on in the development community, in Reddit as well. So I’m limited when you’re adding your knowledge sources. I am only unlimited to 4 external knowledge sources when it’s going to these websites. So again, if you’re looking at driving a high level competitive analysis engine that is going at maybe websites where you have your company has subscriptions to where you need to log in. And understand how you can mine that data or use that data. You have other things you’re connecting up to data, maybe larger data sets, Azure fabric, things like that. That’s where we’re going to use copilot studio. But again, this is going off of what I have available up in SharePoint. So you can see here it came up with a list of suggested prompts. These are not really. They’re all right to get started. So we’re going to start with. Just a question about app categories. What app categories show high revenue per per user but have consistently poor user reviews? That’s going to go through our data side, understand that. Sorry, I was just looking at this other chat here on some it looks like somebody has an AI agent that joined on their behalf today. Let’s see. So it’s going through this large data set again this first time and now it’s going through. I wanted to understand what I have going on in this data set. So categories with high revenue per user and poor reviews, so educational apps and giving me an example revenue models, subscription based targeted targeting parents. Children user complaints. This is pulled from the Google user reviews I pulled on this information app crashes. Poor customer support, slow performance and list glitches. Overall sentiment despite some positive feedback, many reviews expressed frustration over usability and reliability. So this kind of information gives me maybe as a somebody planning on what mobile app am I going to develop next or a company going to develop next. This gives me some competitive analysis. So I understand what other companies are doing well, what they’re not doing well. And maybe I can go out and make something similar that’s going to perform better. So it’s actually looked at all of this information and why these categories matter. These are profitable due to strong monetization models, higher user engagement and they’re underperforming and user satisfaction from technical issues. Terrible UX or design choices and lack of responsive support, so we’re going to go ahead and we’re going to create this agent. And I have some additional questions while we’re waiting for that to load somewhere. That we can go ahead and take a look at. So just a moment. We’re gonna go to our agent. Go back here. Too many windows opening here. So now it’s giving me some of this. So the reason I want to dive into this a little bit deeper is as I start asking more questions, let’s go back and make sure I turned on one additional feature. But I can edit this. One thing I did forget to do was to make sure that my turned on code interpreter so it can actually analyze and generate code and I want to run on the image generator so that I can generate charts and information about my data set instead of just having text conversations. So those 2 toggles there, we wouldn’t use that say for the HR chatbot, we wouldn’t be wanting to do things like that. But in this scenario, if I start uploading code or actually want to start going down those paths as well or I want to actually generate charts about the data that I’m mining, that’s that’s how we turn that on. So we can actually look. And now we’re going to. Go back. We’re just going to start a little bit here, which training categories have simple apps that can be improved with better execution. And as we go through and start diving into the data, it’s going to allow us to start using some predictive modeling capabilities and it’s pulling in some of those Excel advanced analysis capabilities. Directly in copilot for us or in our chat agent. So while we’re waiting for that, I’m just going to see if there are more questions here. Here we go. Do we have any white papers on agent management and governance best practices? There are some that I found that are published out there and we do have a set of those that should be published based on this new dashboard in the month of October. Work. So as those white papers as we get those out, we will share those with everybody. And but you should see those in the next 2 to 3 weeks as we dive deeper into the new features that Microsoft is unveiling at a tenant level for managing these agents across the enterprise. We do have some publications out there on governance, best practices, some different articles on LinkedIn, written by many of our architects, and there is some information out there as well, but we can direct you to some after the call here. Go back, chat. So here we can see again. So top opportunities for simple apps with poor execution and it’s giving me the actual names of the apps and I don’t want to go up against Fisher Price. So you know, even though that app rating is terrible. I I’m not going to go up against a large company like that, but we’re looking at opportunity scores is calculating these for for these for me and showing themes across these too many ads, poor UX, missing core features, performance issues and privacy concerns. So now as we start allowing it to use our default prompts, it’s going to start getting into the data behind this and we can start asking some better questions about how we monetize, how we design and how we build something out. So we’ll go through and it can actually as and once we pick an app, it’s actually going to help us scaffold that and create the UX and it can help us do some different things from code generation there as well. So here we’ve generated now. We have a list again tools, health and fitness. We have this list. Now I would like a. Monetation Strategy MVP, let’s do an MVP feature list for health and fitness. Even though there’s a lot of that out there, we’re gonna let it go that way. So it’s going to create an MVP features list based on overall user sentiment and other apps that are out there based on reviews. What they’re not doing well and what users are seeing that is being done well. So the reason for doing this is we are taking a data set. And we can do this in our own companies, right? We have products that we sell, we have services that we offer. We we all have competitors and are looking at what different organizations are offering. So how can we use this other than you know, Microsoft does have a researcher tool and there is an analyst tool. But when we’re looking at niche markets and we want to bring in our competitive intelligence from maybe services we subscribe to or different things inside the organization, you know we’re only allowed you know, so many external, but this gives us a different targeted approach instead of using this generalized researcher, right. Have tell it every time I do it. Go to the SharePoint site and look at this. Look at this. Look at that. I’m building out a repeatable competitive intelligence robot for mobile application development for my company. So here we have some core features you know, so step counters. We look at tracking. Guided workouts, meal planning, privacy. You know, Monetation ready freemium models and let’s do a show wireframe idea for this app. So we’ve turned on our image generation and we can start using this too. It’s going to give us a wireframe here. It’s going to create an image for us so we can start working with this and start turning this into code and bring it over and start defining what our framework looks like. So this is a really unique way to start using copilot. To take images even and start going OK, I want my user interface to look like this. I want to make this change on this website. How do I do that? I can start doing this but now by adding screenshots or adding different types of information saying OK I want to make it look like this. What do I do? So just different ways of being able to use these agents and this isn’t maybe something I would share with my department, but I’m also highlighting this because this shows how I can use copilot to create a personal productivity agent that’s going to automate or make part of my job easier. So it’s not just about creating agents to go out to everybody in the organization. We can also create agents that help us get better at our roles and help me offload some of the mundane tasks in my day to day to something like this. So image generation does take some time. It looks like it might have not finished so. Ohh, I interrupted and I shouldn’t have done that. It is working on the response. So it’s giving us some very basic wireframes here. It does take a while to generate images when you do them this way, but I can start having conversations and say OK now I will. That’s the activity tracker that I want. Now how do I write that? How do I write that in code? And what does that look like? What is my database structure so we can start changing and turning our conversation and start actually generating code to help us do this if that’s what we’re going for? Or we can take a different direction depending on my role and I can start getting into more revenue models and MVP and feature lists. And building out my user stories before we actually start building out an actual interface. So I think I am going to stop waiting for this image is going pretty slow, so I don’t think we want to watch the slow refresh of an image today. So I’m going to go ahead and go to my next agent while that one is working. So another agent that I’ve created is for customer churn analysis and this is taking a data set where it’s looking at my customers and what did they what services do they purchase for me? How do they pay? Are they subscription based? You know what are all those different services that they are using? How long have they been a customer and then start making predictions about my customer base and where I have churned where I potentially have to watch and different things like that. So I’m going to go grab that data set. There it is. That’s why I couldn’t see it. Press a charity s. Any questions while we’re starting building this next agent, I know we have about 10 minutes left, so I want to make sure if you do have any questions that we get those address while we’re watching progress bars or things like that. OK, I’m going to go ahead. I’m going to skip over to configure my agent. I have a very large data set that has a a, not a customer analysis that’s out there on SharePoint. I’m prioritizing my knowledge source again. OK. Looks like it out of that twice. There’s just a delay in the interface there so. Alright, so we have a customer turn analysis hooked in and now I’m going to go back and I’ve already generated a prompt for this so that we have something a little bit different here. So our churn analysis and this is a data set that is again really large and it’s going to do an analysis of our data set. So it’s examining the data set, identifying churn indicators, reprocessing that data, understanding our demographics, analyzing the services that are. Being consumed doing a financial analyst analysis. Excuse me. So we understand what’s going on with potential subscriptions and churn and purchases statistics and there’s a lot of information here. And then key metrics to calculate and report. Churn rate, churn rate by segments, customer lifetime value, monthly revenue loss amount, percentage model accuracy top 5 most important churn indicators. So I’m going to go ahead and bring this all over and copile it as well. And I’m going to bring this into my instructions. This one is just about 6000 characters again. And it is actually pre populated some pretty probably. Important questions what customers are must get risk of training this border? And then after it answers the first question correctly. I reviewed your data. And it’s giving me the key indicators of an at risk customer. So the majority of churn customers are month to month my payment methods, a significant number of customers, these electronic checks, which correlates with a higher churn compared to automatic payments. Of course, I’m I think we’re all guilty of that sometimes as subscriptions of course. Higher churn rates, even though then mail checks, which I find very interesting checks have a higher churn rate than the mail checks, and that could give us interesting information about the demographics of people that are actually who mails a check. So we can look at our database and figure information about that churn. Customers have a low tenure. So if they’re new, our newer customers are most likely to leave customers with high monthly charges about $90 and more likely to leave, especially when combined with a low tenure churn. Customers often lack value added services such as online security, tech support or streaming services. And Internet type also plays an indicator customers using fiber optic and Internet show higher churn rates than those using DSL. So we can make some potential profiles and understand the high risk profiles in our organization. So again, here’s some anonymized examples of customers who churned and matched high risk profiles. So we kind of look at what our demographics are and we can use this then to start creating content. Do we have to create new marketing content? Do we have to offer incentives, do we have to offer? Something to our highest risk customers to make sure they make it over those thresholds where they start to turn into longer range customers. So here I never gave it a name and I always do this, but I want to make sure I actually allowed it to create data. I usually forget to slide these 2. When I’m doing dumbbells. Go ahead and update. Go to our agent. So since we’ve already looked at our churn risk, I’m actually going to look at retention strategies. So suggest retention strategies to reduce churn for our premium customers. So now we can start looking at different ways of using our data set. And this could be something I could service to our marketing. I could service to marketing, I could service to sales I could service to profit product development. So again it’s going support strategies, it’s going key observations first and it is repeating some of the things we saw during our testing and now it’s giving me. Recommended strategies incentivize with long term contracts. We have some targeted retention for electronic check users, encouraging migration of automatic automatic payments with one time credits or loyalty points, bundling high value services. And predictive alerts for at risk customers. And so here we can go ahead and ask it to create a dashboard for us with churn insights and it’s going to go through and use pandas to analyze our data, use some different code. And now we’re going to see both depending on the chart it puts out. We should see some code as well as an interactive chart that we can actually see in paste and other. All their apps. So just a couple examples today of how we create agents to surface and ask questions across the organization, like a force multiplier. So a lot of those questions you get consistently across the organization, you can offload those different to different agents. If you are looking at augmenting how you do your job, we can do that a little bit differently than just your copilot. I don’t have to manually upload 34510 of the same documents every time I do a certain thing. Whether that’s getting ready for a weekly meeting or a status update. Or project updates I can create agents that are acting on my SharePoint repositories to start servicing information faster, so again offloading parts of my day that are kind of tedious. So I’ve showed some additional copilot agent use cases. We have additional SharePoint use cases in this deck as well for you some different scenarios and then I’ve actually showed some other ones that we haven’t covered today, but policies, onboarding, contract review and these are all made in a way where you can copy and paste. The prompts are here so you can get started on in different. Foreign areas here, so you have the exact prompts that you would use to get started with each one of these, and you can see here there are a lot here throughout this deck. So we do have all those available as far as next steps. If any of you would like to discuss these more we do. Our technical architect can discuss, you know, copilot adoption agents with your organization and exploratory setting. We do offer our executive AI and visioning sessions so we can bring our experts to your office or online. And we can also bring the event to you. So any topic or. Topics from our events, we can actually absolutely bring to your organization. So if there are no other questions, thank you everybody for your time today and we will be sharing the deck after this. Absolutely thanks everybody.