Transcription
We are doing so much work these days on co-pilot agents and AI agents, and Danny, he put together this really, really awesome "Day in the Life of Financial Services" here's the demo flow. We are NextGen Investments, and what's happening is there's a customer called Lily. Lily has been calling on the phone and having conversations that have been recorded and transcribed by co-pilot. She's been asking about like making changes to her investment. She sent an email, Lily, to a junior associate, which is Danny. Danny is kind of a new hire. I'm a new hire, um, I'm a, a junior or associate financial advisor, and I'm leveraging the power of my Financial Advisor Pro agent.
Few things here to show you. Um, you don't need to be, or I don't need to be a financial, I mean, a prompter engineer. Um, the agent already comes with pretty fun prompts. So the first thing I need to understand is what's the latest, uh, in terms of what's the latest from our customer. And then I'm going to change here the name. Lily is my customer, and here's her, her account number in our organization. So K knows now, it's smart enough, my agent, to go to places inside our work data connectors, etc., to pull the latest everything, but that we know about Lily in terms of what's the latest. So I'm going to let her run, and this is all I need to do. Input the name, let my Financial Advisor Pro come up with an analysis of everything. What's going on with Lily? And in this case, it said, "Okay, uh, looks like she had an income increase right here. Um, she purchased a house in the past with 3.5% interest rate, which is awesome. Family expansion, uh, she's having a, she's going to have a son or child soon. And then investment interest. Lily is interested in diversifying her investment into stocks, mutual funds, and real estate, particularly in the technology and renewable energy sectors."
Great. But now I know, we know exactly what's going on right now. What I'm going to do is, since my manager Shervin wants me to create a proposal for a customer, I'm going to start editing in Pages. So if you think about Pages and co-pilot, this is basically a new canvas for collaboration. I'm using the responses created by co-pilot using generative AI and a smart sources of information, and I'm going to be adding those responses into a document, which is what you can see on the right-hand side here. Again, it's going to leverage the sensitivity labels where this response came from. It's going to start adding this in a loop component that I can start using.
So now I'm going to go back and say, "View prompts." And now I'm going to say, "Okay, I know, and because Shervin told me, my manager, that Lily is getting a proposal from a competitor, that competitor is Koso LTD." So I have another prompt here that says, "Provide a competitive analysis for our current NextGen Investment Portfolio with competitor." And I'm just going to type the name of the competitor, Koso LTD, and let it run. I didn't have to write the prompt. I'm just using a prompt that's already there in my agent, and I'm going to have a side-by-side comparison between us and our competitor. Now, it literally takes less than 10 seconds. Co-pilot knows where to go, creates a very nice table comparing our index funds with our, our competitor index funds, and it offers me the sections that I asked: expense ratio, performance, and annual gains. And there's also a column here that says explanation. "Both portfolios have similar expense ratios. NextGen focuses on stability and tech growth, which is what she asked us in that conversation. She's interested in the technology, uh, index funds, while Koso emphasizes in emerging markets." Bingo. This is a good win for us. There's a detailed comparison, expense ratio, why we're better than the competitor, performance side-by-side, annual gains side-by-side. Wonderful. I didn't have to make a marketing competitive analysis. I can go with this.
I'm going to click on "Editing Pages" and take a, and just pay attention to what happens on the right side of the screen. So I'm going to see Pages edited. And now this second section about, um, you remember the first section that I added, the current status? There we go. It's there. And now I added a new, new section about competitive analysis. Here's the table. Wonderful. And here's the explanation. Everything in the canvas, everything in the loop component.
Now, the last thing I want to do as a junior advisor is to go, go back again into "View prompts" and click on this. This is where, this is where kind of the money goes when it comes to you as an organization showing something to your customer. "High Financial Advisor Pro, can you help me recommend the best investment portfolio for customer name Lily with the customer ID based on her NextGen Index Funds overview?" And then I said, "Provide the results of the options in the table with a column explaining why this option was selected. Make sure to rank the offers from the most suitable to less suitable to the customer based on current situation." I'm going to hit enter here. Boom. And again, we can count it. It's going to take less than 10 seconds. Co-pilot finds that information for me and does exactly I said. Creates a table, offers our, our current investment portfolio, tells me why you're putting this on the top, and put this in the table. So I'm offering her five different, five different index funds. You can see that the rank is number one. This is the one that the first one we should offer, which is the Offa Stability Index Fund, and there's a description, there's expense ratio, annual gains. And this is why: "Ideal for Lily's preference for low-risk investments." I did not ask in the prompt, "Make sure you rank low-risk investments as a first off." I did not ask that. This is contextual understanding. In vector search technology, co-pilot knows that Lily asked specifically for low investment risk. She's having a kid, she just purchased a house, she wants to be conservative. Creates a table, put them organized here. There's a reason for that selection. There's annual gains, expense ratio, description, funding, rank, and there's an whole full analysis here that you can read, validate, understand. Good to go. Um, beautiful. Again, I'm going to click and "Edit Pages." Take a look at the right-hand side of the screen. That piece of information will get added into my document here. Wonderful. This is getting shape and it's looking like a very nice proposal that we can send to our customer.
But we're not going to do that just yet. Well, Danny, I just want to say as your manager, you are an amazing junior agent. I mean, look, look, look at this body of work that, you know, you're just a new hire into the company, and because of these tools, you're able to hit the ground running and produce incredible and very detailed analysis. Right? And just to kind of summarize what you did was, first, you did a discovery of what the customer's ask was, right? And it was, that was the one prompt that you just clicked on from the agent, and it, it, you know, it went and did all the hard work for you. Two, you did a competitive analysis because that's what the customer was saying they were comparing, you know, our funds with somebody else's. And then three, you asked for like, what are the funds that we can recommend based on the portfolio that we have? So those three things were basically just prompts that you clicked on, and co-pilot did all the heavy lifting, and you put it inside Pages, and you documented it. So, I mean, incredible work. And this is the power of AI agents. And there are different ways to see the value here. One, you have a new hire, uh, that's learning and understanding how this works. They're going to get up to speed quickly. They can understand how the business works. They can understand about all these terms. And I don't know if you noticed that the sensitivity label changed from "General" when I run the first prompt to "Confidential." Microsoft extended why? Well, that's because one of the responses that I got from co-pilot came from confidential documentation. As soon as this information comes, by the way, this is a demo, this is not really confidential information, but the point here is, if the responses from Microsoft, from your prompt, come from confidential information, it's because one, you had access to that information, otherwise co-pilot would never give you the answer. One, two, as soon as I added that piece of the response into my collaboration canvas, which is called Pages, the sensitivity label of my Pages changed right away from "General" to "Confidential."
So now that I'm ready with my proposal, I'm going to share this with my manager for approval, for review, before we share with the customer, right, with Lily. So I'm going to click "Share" and I'm going to say "Copy component." I'm going to click in "Settings" and I'm going to "People you choose," and I'm going to send this to Shervin, here's my manager. So there we go. Click on it and apply. Okay.
So now that I have shared my loop component with Shervin, so we can start collaborating in in the final proposal. Let me show you how that looks like in the web, just because we have more real estate to work with. I can see that my manager Shervin is now part of The Proposal. I can see that he's in, as you can see that he's moving the mouse, he's moving around the document, he's making changes. Can we add a table? Okay, great. So he's actually asking me to do a few things here and there. If you also notice on my screen, I can click on co-pilot and then start rewriting these pieces of this information with co-pilot. That can make changes. I do want to ask Shervin, since he talked to Lily about, I want to ask about this particular section about family expansion. So I'm going to create a comment here, and I'm going to ask Shervin. I'm going to @ Shervin, um, "Please, please confirm this is accurate." My manager is coming here. He says he's looking great. Finally, Shervin thinks that this is a great document. He's going to approve it to me. So then we're ready to, there we go. I got a boost from my manager. Wonderful.
So that, like, so this is the document that we're going to be presenting to, to our customer. The next thing I want to do now is I'm going to go here and I'm going to "Print" and "PDF export." I'm going to create a PDF document out of this particular loop component that has all the information coming from co-pilot. You can see you can make all the changes that you want. I'm going to go ahead here and print this document.
Finally, I want to show you how you can create a PowerPoint deck that you're going to be presenting to your customer using co-pilot in PowerPoint. So I'm going to use a predefined, pre-approved, uh, template. This is the one from Microsoft 365. So I'm going to go click here. This is a, a template. As you can see, fonts and colors are all there. And I'm going to click on co-pilot and I'm going to go into the "Narrative Builder." Co-pilot here, representation about this particular file. Again, this is the Lily Lives customer proposal, and I'm going to run it. There are different sections. They create a section for the summary of the latest communications, key points from the meeting, investment interest, comparison with Koso. So fantastic. It's covering everything. Let co-pilot generate the slides.
Finally, once I have this PowerPoint, which is ready to go, um, I'm able to schedule a call with the customer, Lily, and come up with a proposal. So she's going to see a folder, a PowerPoint that has speaker notes to help you present that information. There's an agenda, there's a current status where she is today, investment interest, this is what she's looking for, what about the income increase, particular real estate purchase, and all the things that we collected from, um, our particular conversation. We're now ready to go. Again, in a matter of minutes, this is where co-pilot generative AI can really help you boost productivity and customer satisfaction. Congratulations. You know, I'm going to give you more of a workload now. Like, if you can bang out these in like 10 minutes, you need to work harder. So you're too efficient. Then yes, that's right. So I need a raise right now. I, I wouldn't mind. And, but by the way, Danny, why don't we show like this agent that you created? You know, like now that we've showed this demo, maybe just a sneak peek into the agent itself. So whenever you get a co-Microsoft 365 co-pilot license, within co-pilot chat, you can see where it says "Create an agent." I'm going to go here to "Create an agent," and then finally, I'm going to go over here, "View all my agents." So I'm going to show you the Financial Advisor Pro. I'm going to edit so you can see how this was created. Again, it's a low-code, no-code. You can see that it's using a lighter version of co-pilot studio in the back end. I'm able to add an icon to it or an avatar, name, description. There's, uh, some instructions here. And here's the key, right? This is probably the most important thing about this co-pilot agent is the knowledge. This is where I can start at telling my co-pilot agent, these are the things that you need to go to pull data for my, for my purpose, right? In this case, I have a SharePoint site, I call it FSI demo. It can be any SharePoint site in your organization. You can enable the web access if that's what you want to do. And then here's where you can start adding connectors. These are what we call extensibility options. Connecting to a Service Now instance, maybe connecting to a wiki page, connecting to an SQL database, connecting to whatever you want. Again, sky's the limit. If there's an API, we can get there.
So Danny, thank you very much, man. This, this was amazing. This was awesome. You are probably, you are going to get a raise. I'll tell you that much because, you know, even though you just got hired in the company, that's my junior financial advisor. You're doing such great work that, you know, I can't help but promote you right away, right? So you've done fantastic work. This was a great demo, great usage of AI and co-pilot and agents. So thanks again. And, uh, I'll leave the last words with you as well.
No, no, I want to thank you for the opportunity to come here, join your channel, and explain, kind of help your community, uh, understand and learn more about co-pilot. You have a great community of users. So, uh, we want them to learn more, understand more about this. Reach out to your Microsoft representative if you want to learn more. Uh, what we showed you today is a financial services use case, which is what I do, kind of, on a daily basis. But sky's the limit here. Like, you can do anything, everything you want. And with the, what's coming next now is autonomous agents, which I know Shervin and I are working on something internally at Microsoft too. Yeah, to help.
And I think on that point, and, you know, you're wearing such a cool hat that I got to put my co-pilot hat on too. You know, I can't be outdone with that cool hat. So I was just going to say that there's so much more to come. You know, we're just scratching the surface here. So whether it's agents that are using connectors and APIs, or autonomous agents that have triggers that are automatically being used based on emails or like documents or new users being added, they're just kind of running in the background. We're going to be building these things and doing more content for you all. So if you are interested in anything specific, let us know. But Danny, again, man, thanks so much. And you like the hat? I like it. You should send me one of those. I don't have that. Well, I got, I do have a bottle. Oh, the bottle? Maybe I'll trade you for the bottle, man. Well, we'll see. We'll see. All right. Okay. Thank you. Thanks, Shervin. Thanks. Thanks, everybody. Take care. Bye-bye.