1. 73, 74, 78. Uses and where you can publish chatbots, creating our first chatbot
Hello, and in this section of the course we’ll be talking about Power Virtual Agents. And you can see on the screen the sort of thing that agents are about. They allow you to talk to a machine that might be able to give you information. For example, are there any stores around me? Well, where do you want to know about Kirkland? Well, our place is here. And did that answer your question? Now, why would you want to use a Power Virtual Agent? Well, you can solve fairly common customer issues automatically, and this frees you and your staff up to handle more complicated requests. So if you answered no, did this answer your question? The next thing could be that I’m now connecting you to a human who can help. And not only can you solve customer issues automatically, you might be able to solve employee issues automatically as well.
Now you can use chatbots on websites, mobile apps, Microsoft Teams, Facebook, and other messaging platforms. You can also have authentication with your chatbots as well. For example, as your active directory, as your ad, Microsoft Accounts, or Facebook. They can all be used to authenticate Power virtual agents. So what we’re going to do is sign in and login into Power Virtual Agents. So you can see the website that I mentioned, powervirtualagents.microsoft.com, and you can get there by just googling “Power virtual chatbot.” So let’s click on “Start free.” So now we have to sign in using a Microsoft sign-in agent. And the first time I tried this, it said that it couldn’t sign in with some badly punctuated sentences, and in fact, I seem to be having multiple problems with multiple domains. Maybe there’s a problem that Microsoft sent, but eventually I was able to sign in using the address that I got when signing up for Microsoft Office. So let’s click “next.” And there we go. All set. So let’s get started. And here we are with powerful virtual agents. So do I want to receive information, tips, and offers about Power virtual agents? Yes, I do.
So now it’s asking me to create my first bot, and I’m going to do that. I’m going to call it my first bot in the language of English. You can see all the languages onscreen that they support, plus preview languages that they are developing and my environment. So let’s click “Create,” and I’m speeding this up quite a few times. It does take a bit of a while to log in and create your bot, but eventually, here we are. So I’ve created my first bot. If I wanted to create a subsequent bot, I could go up here in the title bar, where I’ve got this bots panel, and I could create a new bot there; if I wanted to delete it, I could go to Settings, general settings, and delete the bot. Right. So what we’ve done in this video is look at Power Virtual Agent use cases. Why would we be using PowerVirtual agents for the first time? Solving common problems and freeing up your staff for the more complicated stuff We’ve discussed where we can publish them: websites, mobile apps, Microsoft Teams, Facebook, and other messaging platforms. And you can use authentication. And we have signed up and created a chatbot.
2. 81. Test a chatbot
In the previous video, we created our first chat bot. So let’s go and give it a test of it.So you can see I’m in the home section, and we have this test bot at the top left-hand side. Now, I can remove it by clicking the X, or I can hide the button in the bottom left corner. So I’m going to use the sample topics that are included in this chat bot. And I’m going to ask the computer. Well, actually, I’m going to say hello to start with. ith. HellI’m a virtual agent. I can help with lots of things. If you want to speak to a human agent, please let me know. So what can I help you with today?
So, when are you going to be open? And the answer is, I’m happy to help with store hours. We’re open at this time. Did that answer your question? No, I’m going to say, “Sorry, I wasn’t able to help better.” Would you like to talk to a human being or try rephrasing? Well, I will try rephrasing. And I’m going to now say, “Where is a store near me?” What location am I interested in? Now I only have three stores. one in Redmond, Seattle, and Kirkland. Well, I’m interested in the one in Kirkland. So I’ll click on that, and it says, “Okay, that’s where we are.” Does that answer your question? Okay, so I’m going to rephrase again, and I’m going to say, “I want to buy items.” Okay. What state will I be shipping to? Well, I’m going to the state of New York. There will be a $27.50 additional charge. Is that acceptable? Yes, that’s quite acceptable. So I’ll click on yes. Oh, what am I interested in purchasing? Well, a desktop computer, a laptop computer, or a gaming computer? Well, I think it’s a gaming computer.
Go get them, tiger. I’m adding this to my cart. Obviously, there isn’t a cart that they’re actually adding to. Did that answer your question? Yes, it did. Great. Please rate your experience. Well, I will rate this five out of five. Can I help with anything else? No, thank you. Thanks for chatting with me. Bye. Now. That was a bit of an artificial conversation. I didn’t really want to know when they were open, and then where were they? in den by actums. But it was interesting to see what the response was and, more importantly, how it was created. You noticed that as I was going through, I was clicking on some of these answers that I was being given, and you’ll see that we are being driven down particular conversation routes, some of them fairly simple, some of them a bit more complicated. In the next video, we’ll be looking behind the scenes at how this chatbot was created.
3. 75, 77, 80. Topics, entities and actions
In this video, we’re going to take a look behind the scenes of Power’s virtual agents. The first thing I wanted to talk about were topics. Topics are conversation plans. So I can have a conversation with a very simple plan. Here is a conversation about something very small. It is: when are you open? Or when are you closed? or that sort of thing. Trigger phrases are now used to start conversations. There are other ways that it can start, but this is the key one. Trigger phrases are questions or phrases that are related to a problem. So when are you closed? When are you open? What are the store hours? What are your daily opening hours? They all go towards the same thing. Now notice I don’t literally have to type exactly those things. For example, on another topic, the trigger phrase was “buy items.” And I said I wanted to buy items. The computer uses artificial intelligence to pass on your customers’ input. In other words, this is the entirety of it.
What does it actually mean? How does it relate to any of these key phrases? Now, I’ve not actually added any of these topics myself. When you create a chatbot, you have these four user topics as lessons, and you have some system topics that you probably will need in conversations. Greeting escalate. That means talk to an agent; the conversation is over. confirm success or failure. Goodbye, start over, and thank you. Now these system topics cannot be deleted, but they can be modified. However, these lessons can be deleted. Now the next thing I want to talk about are entities. And unlike the entities that we’ve got in the database, which are now called tables, these entities probably won’t be renamed as tables because they’re not anything to do with tables. They are either real-world subjects or concepts. For example, there is no real-world subject for a telephone number, but we can get telephone numbers from conversations, so please call me; the entity will be the telephone number. Or alternatively, you might have zip or postal codes or addresses, cities, states, colors, numbers, temperatures, or even names.
So, if you’re passing a sentence, or reducing a sentence to its constituent parts, you might need to extract things like temperatures. If somebody reports that their oven is too hot, it won’t go to 220 degrees, 350 degrees Fahrenheit, natural gas, or whatever you use. So this can extract parts of text that you can use as part of your conversation. The third thing I want to talk about is actions, which we already know about. Actions allow chatbots to call a Microsoft PowerShell automation flaw. So they have to be in the same Microsoft Dataverse environment as your chatbot. Floors must be in a solution in Power Automate, and floors typically use variables for input and output data. So let’s say I was calling a PowerAutomate floor and I wanted to bring along language or email or something like that. While those were variable, they would be used to power the automate floor. So in this video, we’ve had a look at topics that are real-world subjects or concepts. We’ve touched on the fact that you can trigger a Power Automate floor with actions, and we’ve got topics that are conversation structures. In the next video, we will go into these topics and see what these structures are all about.
4. Looking at topic conversation flowcharts
So, let’s take a closer look at some of these topics so we can see what kind of things we can have. First of all, I’m going to click on Lesson 1, a simple topic. This takes us to the trigger phrases. As a result, these trigger phrases are the start of a topic. Someone says, “When are you closed?” This is when you go to this particular topic. So if I now click on Go to Authoring Canvas, You can see this is a fairly straight structure. We start off with the trigger phrase, have two messages, and then end the conversation. And you remember when we ended the conversation? It said, “Did that answer your question?” Yes. No. So here are the messages from the topic: Let’s go to a more complicated topic. And then ask, “Are there any stalls around here?” Now, there are three stores for this particular company, and they are in Redmond, Seattle, and Kirkland.
So they were presented to us as multiple choices. And we went to a condition based on what you said. So if it’s equal to Redmond, give the information about Redmond; similarly for Seattle and Kirkland. And you can see that we are saving the response in a variable called store location. So we can say if the store location is equal to Redmond, then our store location is located at the end of the conversation. This lesson three is a little more difficult. It asked me where I was going from; what is my state? and I said New York. I deliberately didn’t say California or Washington. If I had said that, it would have said, “Okay, no problem, the shipping rate is equal to this.” Shall we continue? Otherwise, if it wasn’t equal to Washington, Oregon, then we’ve got an additional shipping charge of $27.50. Is that acceptable? And if we said no, then thank you and please come again. Otherwise, we set the variable shipping rate as being equal to true. Ask what you are interested in buying. A gaming computer, I said. And so we get the messages regarding the gaming computer, and then I add it to your cart. Please go back to your cart.
Now, you’ll notice that all of the options that I’ve selected or that were consequences of what I selected have these check marks and a green background. So you can see how I’ve actually gone through this particular sequence. Now, we also have a button on the top here to “track between topics.” So this allows me to see—okay, I’m here. And if I were doing this in real time, this would fill out even if I moved to a different topic. Now I have the option to continue the conversation. And for instance, if I say, “When are you open?” It will now trigger the first conversation. It’s now gone to the end of the conversation. So did that answer your question? The term “electronic commerce” refers to the sale of electronic goods. Okay, next conversation. Where are your stores? I’ve got to click “rephrase” just to get onto that. So where are your stores? So it gets me onto another topic, though it would do if I were actually actually getting the exact wording that they wanted or something close enough with artificial intelligence. So now it’s taking me down here, and I’m going to say that this time I’m interested in Redmond. So you can see that the track between topics allows you to see the latest topic that it is currently on. In the next video, we’re going to create a simple topic.
5. 76, 79. Creating a topic, using nodes, conditions, and trigger phrases
In this video, we’re going to create a simple topic of our own, so I’m going to click on topics and go to new topics. Incidentally, there is a suggested topics page that we can also go on; it can look at your online data and give you topic suggestions, so let’s start afresh with a new topic. So this new topic is going to be music and what tastes a person has in music. Consequently, the shorter the trigger phrase, the better. So I want to listen to music and that sort of thing. It says to try to start with five to ten diverse phrases, maybe the word “classical” or “rock” or something like that. So now we’ve got these trigger phrases. I’ll go to the offering canvas. You’ll notice this has absolutely nothing to do with a store, but it might just be helpful to personalise the bot. So what sort of music? No, I like listening to music too. So this is a message.
So this doesn’t allow for any input or anything like that. Notice you can have bold italic, you can have a bulleted or numbered list, and you can also have a link, which might be nice to say, Have you thought about going to a music store and having a link to Amazon, for instance? Next, I’m going to ask a question, so my question is, “What sort of music do you like?” And as you can see, there are various types of choices. Multiple-choice options are obviously the easiest, but you’ve also got entities built in. So, if you say what the temperature is like, you might have the temperature entity to use in subsequent steps. So I’m just going to do multiple-choice options because I think this is the easiest one. And I’m going to say classical, rock, and pop. So, if the VAR text has a better name than the VA music type, I’ll use that. If my music type is equal to classical, then I want to display a message saying I love Betovan. If it’s equal to rock, then you can say, “I don’t know much rock music, I’m afraid.” And I think we’ll show this person’s age. If it’s equal to pop, then have you heard about the Spice Girls? 20 or two years ago, but it doesn’t really matter.
So hopefully you can see how we plan the topic discussion. So what I can do next is ask another question; I can add another condition, so maybe follow up by saying, “Do you prefer Beethoven maltza?” That sort of thing I can call an action. So this will allow me to create a new flow in Power Flow. Or maybe I can manage authentications. I can show another message. I can go to another topic. So some of these topics are end topics or start topics. Or I can end the conversation. I can end with a survey so we can see how successful this conversation was. Or I can transfer to an agent. So I’m going to end the conversation with a survey, and I’m going to end this conversation with a survey. We’re starting to lose our footing now. So what I’m going to do is delete these two, and it doesn’t show how you can actually merge them back. So we separated out these conditions. What you do is click on the plus sign, and then there’s a little dot that’s here, and you drag the dot to another dot. So these two, and now three, separate parts of the conversation are reintroduced into the same stream. So there’s my conversation. I’m now going to save this. Now I can if I want to. For instance, if I wanted to add another message, I could insert it in between two items. So maybe I want to send a message.
Notice my list is a bit more limited. I can’t now end the conversation from here. I can’t jump to another topic from here because this one is continuing. So if I were to jump to another topic, well, the rest of the conversation would never happen. So can I ask you a question? Now, suppose I had something else, like, for instance, I was going to add in a condition, but I didn’t actually finish anything. The topic at checker will allow me to see that there is an error. Now I can save it with an error. So click “Save.” Do you still want to save? But bear in mind that I won’t be able to run it. So it might be that a node is missing, a field is missing, required information, data, or an expression is invalid. Or maybe I’m using a variable like music type. But then I go up here and type the initialization of it. So it doesn’t make any sense. So let’s rectify this error. I will end a conversation with Sarah and link these two back together. There we go. So now let’s test this. I’m going to say, “Ask, when are you open?” Notice the track between topics is on “click save.” I’ll just click save, just in case. So when are you open? And here we are. Did that answer your question? And I’m going to now ask, what music do you like? I miss music too. Have you thought about going to an online store? A music store? Can I ask you a question? What sort of music do you like? While I like classical music and I love Beto, But did that answer your question? Yes. And then, can I write about my experience? I think another five stars is in order. So in this video, what we have done is create a new topic. It’s a very simple topic, but within all of this topic, we’ve got all of the building blocks that we can have. We can show a message, ask a question, add a condition, and then branch onto this. And we also found out how to branch back. We can take action, go to another topic, or end a conversation.
6. 82. Publish a chatbot
In this last video of this session, we’re going to have a look at everything else that’s here in Power Virtual Agents. First of all, the Manage tab We currently manage channels, which just gets you to this list of channels. Just hide this bot’s details. Here are the details of your bot, and you can change the icon. You can also add security to control who can use your bot by, among other things, verifying users’ identities during chat and inviting people to collaborate on your bot. You can enhance your bot so it can book appointments and send confirmation emails, manage tasks, and more.
And then we’ve got analytics. So we have, first of all, the Summary Table of Contents page, and I’ve seen it referred to as both. So this gives you an overview of performance with charts for engagement over time, session outcomes, resolution, escalation, and abandoned rates. So was your problem resolved? Was it necessary to escalate it to a non-robot? Or was it just abandoned? We also have a number of charts at the top. Now the default is to show these indicators for the last seven days. This can be changed at the top of the page. The Customer Satisfaction tab gives you topics, engage sessions, resolution, abandoned rates, escalation rates, and the average CSAT score. So, on a scale of one to five, how are you doing? This is both the customer satisfaction score and the impact score. How are you doing over time as well, sessions? This allows you to download your transcript. So, in other words, what’s actually happened? I said this, and the bot says this.
Now, from both Summary and Customer Satisfaction, you can drill down into various topics if you want to. So I could dive into lesson two. For instance, you’ll notice that as soon as I click on this, this Power-Bi symbol comes up. So this shows that this is all powered by Power Bi.So if you have some skill in Power Bi—and I know you have by now in this course—you’ll be able to interact with this a lot more easily. And then, finally, there is billing. This shows you the number of build sessions. Now, you remember in the previous video that I said the price was around $1,000 for 2000 sessions. So what is a session? Well, it starts when a user topic is triggered for the first time. It ends when the chat session has been closed by the user or it’s been more than 60 minutes or 100 turns. A “turn” is an exchange between the user and the bot. And if that happens, the end user won’t see anything different. It would just be a separate session. So again, you will see what it is over the last seven days, and you can change that. And you can see all of this from the left-hand side once you get into analytics. So you got the Summary Customer Satisfaction Sessions, which are the actual text of the sessions, and then how much this is charging you in billing.
