Build a Conversational Bot in a Couple Minutes

Hey there, I'm going to show you how in just a couple minutes you can build a chat experience for your customers that answer the top-of-mind questions. It's called a Q&A or question and answer bot. So let's go in and take a look. Here you'll see I'm on the Q&A Maker page. You'll want to sign in and I've already done that. Then we'll want to create a new knowledge base. Once we create a new knowledge base, or in order to create a new knowledge base, you will have to go in and create a Q&A service in the Azure portal. I've already done this, but you'll want to make sure that you go through the process and do that. Once that's done, you'll just select where in the Azure portal, under which subscription and which tenant you've created this, and I will go here. This is my dev subscription, and then I'll look and you'll see I have a Noelle Asharity demo created. I then want to name it. Now in this case, I'm actually going to create a Build a Bot Assistant. Build a Bot Assist. It's got a nice ring to it, actually. And what I usually encourage people to do is think about the name of your bot, right? The name of your bot is really important. You don't want to name it something like Max or James unless you have a really strong brand to support it. You want to give the user an indication of what it's going to do, how it's going to help them. Once you've identified that, you then need to fill the bot with questions and answers. So in this case, we actually have a couple options. One is that we could use a document that has all these questions and answers in it. Sometimes companies or customers will have Excel spreadsheets full of these questions and answers that we can use to build the dialog. But even more companies have an open on the website FAQ, and that FAQ is perfect. For this type of use. So I'm gonna actually go out and use the Bot Framework's frequently asked questions. So I'm gonna copy this URL and add it into QnA Maker. Then you'll notice it also has this really cool function called chat, chit-chat, and this chit-chat is the ability to add a personality to your bot. So again, this is one of those kind of core principles of building a chat experience. What happens, and so many people do this, what happens when you ask a bot like, how old are you? Or do you like coffee? Or do you sneeze? You want your bot to have kind of a personality to it. Like, is it friendly? Is it fun? Is it academic? Is it professional? Right? And this gives you the ability to add an index full of questions that'll help you solve for that. So I'm gonna use, which I like, 'cause I'm the friendly type, I'm gonna use the friend. Then we can create the knowledge base. Now what's happening is it's going against this document, right, that URL, and it's extracting all of the questions and answers and generating a natural language dialogue. This is using an artificial intelligence, a machine learning model, but the nice thing is, is you don't have to build that model, right? We actually can use a pre-built model to do that work for us. And poof, right now I have a curated list of questions and answers that are now informing my chat agent on the most common questions I know my customers are gonna ask when they come to my website. So you can see here, it lists out all the questions. So another really interesting point is that many times customers that come and use your chatbot, they might ask a question that's not a direct line to an answer that you have. What you can do is learn every time your users use your bot, you can learn from it and come in and edit it right here. Edit it, retrain the model, and it will be able to pick up those nuances as the system evolves. That's the beauty of machine learning, right? You can constantly train the model to become better and better at solving your customer's needs. Now let's go ahead and test it just so you can kind of see how this works, right? So here we know we have a question like, "What is the v4 SDK?" So instead of saying that, I might just say, V4 SDK. And what it's gonna do is try and find that entity within my knowledge base. And you'll see it's able to go in and pull up this answer. Now, I didn't have to go and say, make sure you handle all of these different ways somebody might ask for this, right? This is the beauty of an artificially intelligent natural language engine, right? It is able to make those types of decisions for you. So not only does it provide the ability to, get a direct, natural language way of communicating to the most commonly asked questions, but it also gives the freedom to the user to be able to ask it in any way that they see fit. Now, I want to go in and show you very quickly this chat, chitchat feature. So let's say I just want to say hello to the bot. I come into the bot and I say hello. Now, if you do not do something special, what will end up happening is the bot will come back and be like, I don't understand that. That's a really bad experience for a user who's just trying to play around and get to know what the bot's capable of. So this allows you to see, like, for all of these things that a user might say, testing, can you hear me? Can you hear me now? We can actually create a special response. The response you create here is critical to kind of the persona that you're building for this bot. And it should be something that you think through, something that you are very, deliberate about, right? Because this is going to be, you know, the personality of the bot that you create. There's many bots out there that you can look at to kind of give you a sense of this done well. One really great example built on Azure Bot Framework using the same technology is the Flo Bot, right? So the Flo Bot is, if you know a little bit about the Progressive Insurance Company, they have this character that they've had in their commercials called Flo, and It's given them the ability to then translate that personality and quirkiness into a bot. Goodness. And given them the ability to transition that into a bot. So it's a really great thing. You can go see it. It's on Facebook Messenger. You can play around with it. It'll give you a quote, you know, just a great way though to see a real-life application of this technology working and how it's delighting customers and how you might be able to do the same. So thanks so much for joining me, and I hope you enjoyed it.