Introducing DialogFlow

Loony Corn
A free video tutorial from Loony Corn
An ex-Google, Stanford and Flipkart team
4.2 instructor rating • 73 courses • 131,235 students

Learn more from the full course

Hands-on Chatbots with Google Dialogflow

Formerly known as API.AI

04:08:01 of on-demand video • Updated December 2017

  • Build and deploy natural-sounding chatbots
  • Design and build sound interaction models for your chatbots
  • Deploy web apps to Heroku
  • Develop third-party apps for Slack and other messaging platforms
English [Auto] Building that works for your web application or web site is all the rage nowadays in this course. Let's see how to use one of the cutting edge platforms in track marks. We'll see how we can use dialogue. This course has very little theory. It's completely hands on. Which is why it's called Hands on chat or the Google dialogue. In order to keep learning engaging at the beginning of every video will ask you a question at the end of this video you should be able to confidently answer what the two categories of chat boards are and how they are different. If you can it might help you to go back and revise the with your Eneas. It would have been impossible to believe that any developer of a site or Tobeck app can build a chat board which we conduct an intelligent conversation in a very normal manner with a human being. But with the advances made in machine learning over the recent years and the law of cutting edge platforms which allow you to harness this machine learning have made possible these intelligent parts which can conduct intelligent conversations mimicking a human being advances made in machine learning allow speech recognition making it possible to understand words in different languages and different accents. Machine learning also enables natural language processing enabling a machine to understand what exactly a human being said. But really this isn't enough. If you had to write your own machine learning algorithm in order to build your chat but then truly really trackballs would not be as accessible as they are to be the development of these platforms which abstract away the complex technologies underlying boughts of what neat building these parts simple. And one such platform is Google's dialogue. Google Psylocke floor was formerly known as it used on AI. In fact it changed very recently. And you find that several of the screenshots in this course still reference it. Eli this is an amazing platform with machine learning algorithms powered by Google's technologies and an easy intuitive very simple UI allows you to build sophisticated checkbox if you're picking this course you probably already understand right track parts are needed or where they might be used. We cover that anyway briefly but first we start off by understanding what exactly. Mode if it's basically a computer program that is capable of conducting an intelligent conversation about a set to be intelligent if it can simulate human behavior and possed what is called the yarding past the plodding past was developed by the scientists are building back in 1950. And it basically says that if you can interact and have a conversation with a bot and be unable to distinguish between a both's responses and the responses of an actual human being then that bar has passed the Turing test. Now this seems like a big ask back in 1950 but all the more technologies that are available now this will be Bart that past the plodding past let's see some of the simple use cases where you might need a check. But let's say you're in a very complex site. It might be an e-commerce site a matrimonial site a hookup site and you have some trouble navigating it. If you spend enough time on the site and you seem to be aimlessly browsing you might have noticed that a chat no often pops up I ask you to just ask questions asking questions about what you're looking for. What is it that you hope to achieve and so on. This chat is usually powered by a book that has no real human behind this chat window using the chat window is an easy natural and conversational me for you to achieve your objective on that Web site or web application. Jack bots can also be used to debug or troubleshoot issues. Let's say you've placed an order for a pair of shoes from your favorite e-commerce site or the order hasn't arrived yet and you haven't received status updates before tugboats became ubiquitous what you typically do is to call up the customer support agent on the phone and say that you're looking to track down your order. This proves to be very expensive for the e-commerce site because having these thousands of customer support agents and having them on the phone is an expensive setup. If you're in charge of building and running this e-commerce site you might figure out a good way to cut down on costs would be encourage your customer to use a chat window for the Square-D rather than calling up a customer support agent directly. This chat can be powered by a board which is capable of asking the right questions in a friendly manner and directing the user to the right place. Tracking down the status of his or her order and in general being helpful. And really there are many many examples of how bots can be useful. These are just a few business use cases but you can have chalkboards which help you track down the better the better but it can help you with your groceries give you the latest news help you with your finances and also help you shop what we spoke about so far was about jackpots in general. Now let's talk about Bi-Lo look specifically look Floris Google's offering for you to configure and Jack boards very very easily and be able to harness Google's underlying machine learning technologies in order to have naturally intelligent bot which passes the Turing test. Now trackballs can be divided into two broad categories. The first of these is a rule based Jack but you can think of a rule based Jack barters belonging to the machine learning either these boards would basically look for standard actions in the user requests like or did the user mention all those tactics. Maybe they should look through all of the user's previous orders. At the very basic level you can think of a rule based jackpots as working with a bunch of preprogramed rules a bunch of if then else statement. If you find it easier to imagine a factory it can't learn or understand new concepts these barks and to be static and can change only when they are reprogrammed rule based jackpots are a thing of the past. What we work on nowadays are artificial intelligence based checkmarks. These are intelligent they understand what they use that are saying why our natural language processing. They also learn from previous conversations that users have had such that a chat port which has been running for a long time save for two or three months would be more intelligent and the board which existed at the very beginning. These boards are dynamic and because they constantly learn from human conversations they get better over time. Google's dialogue flow is an artificial intelligence based Jack bot that I Robot idea of mod's in which dialogue flow operates. We see that in a little bit but it's primarily driven by machine learning algorithms to learn from me. Let's take a little walk through by and understand how dialog looking to be back in 2011 a company named speak with LAPKIN intelligent personal assistant for mobile phones this person assistant walked on Windows Android as well as iPhone devices. Now this personal assistant was popular and based on that cutting edge machine learning technologies. And in 2014. Speak to it made it possible for third party developers who use these technologies as when they at least an SDK or basically the API don't I like believe third party US. This allowed anyone to plug into the natural language processing and other more use to build conversational apps because with technology was cool. Google in 2016 realized that speak to it could be used to power Google a. You might be familiar with this assistant on your phone which is in the bill. When you say OK Google this is if you have an Android phone. The EPA wrote it is DKs and libraries are still available for third party developers to use. Now this is how it continued till very recently that is October 2017. This is when yeah it was renamed to dial up. This was the Army. A name change. None of the functionality that it gained or I offered was changed. In fact new functionality is being added at a very fast clip. Now this name change turned out to be a bit of a bother for us because it happened right when we were recording the demos for the scores. So you'll find within the demos the UI says if they don't. But when you log in at this date then you will see dialog flow. Don't worry it's the exact same set up. A few things might have changed because Google is constantly adding new features every 15 days or so but the underlying core remains the theme dialog flow. Is essentially a P.A. AI with a new name. So what will I look for to help us. It will allow us to build natural rich conversations using a very intuitive UI. The UI I must say is Amazing. The UI for dialog flow is so simple that you just need to use it once before you have mastered it. I've also used Amazon's Alexei UI and I must say that dialogue flow is generations ahead we mentioned earlier that dialogue flow is an artificial intelligence beast of BART which means that uses machine learning to understand what is it that the users are seeing. It figures out the intent of the use of conversation or using natural language and other machine learning techniques. What sets dialogue aside from other chat bot platforms that might exist is that every machine learning more at it bills for an engine that you can figure out is unique to that individual users bought that machine learning Molly Lundstrom conversations that your users of your website or app have bought resulting in a bot that is tailored for your unique purpose. What are the other capabilities that dialogue flow possesses. It supports 40 plus languages including French German Spanish etc.. You can have the very same board set up to understand multiple languages if you have users from different countries. You can use 16 different programming languages in order to configure the bot including python C-Sharp java javascript all the common ones dialogue for all of us. The same is the case that if I got you I did allow you to build boards for your own application then that is a mobile application or it's a Web site. Dialog flow also offers as the case which work with variables so it's internet of things enabled. Or if you want to build a track or app which integrate with an existing messaging platform like Facebook why bother with your card. Next up or slack that out one click in English intervenor be with all of these platforms you'll see how easy it is to set up the words. The very end of this course and this is a good time to revisit this question. The two broad categories of art are who'll be struck by an artificial intelligence based attack Bartsch rule based hatbox are all in that you have to reprogram the rules which they understand. They look for specific words or actions within the use that expression and then try to give their responses. They are static and do not change over time. They don't learn from user conversations. Artificial intelligence based jackpots use machine learning and natural language processing in order to understand what the user said. They are more intelligent. They are dynamic. The longer you have the bottom production the more intelligent it will get. The dialog platform helps you build AI based checkerboards.