Microsoft 365 Community Microsoft Teams, 9 Jul 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Using Bot Framework Composer to build Bots without writing code

Description

In this 18-minute demo, Stephan Bisser shows Makers and Developers how they can use a new tool called Microsoft Bot Framework Composer to create a Bot from a template or from scratch and in one environment. Scenario defined for this session: Build a very simple Bot that can answer questions stored in a Q&A Maker knowledge base and implement language understanding. Bring to the table - Bot name, Q&A Knowledgebase ID, and basic understanding of Bot functionality. This PnP Community demo is taken from the General Microsoft 365 Development Special Interest Group (SIG) - Bi-weekly sync call recorded on July 9, 2020.

Demo Presenter(s): Stephan Bisser (Solvion) | @stephanbisser

Supporting materials:
• Bot Framework Composer Documentation | http://aka.ms/bfcomposer
• Bot Framework Composer Series - 1 - Use Adaptive Cards | https://bisser.io/bot-framework-composer-series-1-use-adaptive-cards/
• GitHub - Microsoft Bot Framework Composer | https://github.com/microsoft/BotFramework-Composer
• Microsoft Bot Framework Web Chat | https://github.com/pnp/sp-dev-fx-webparts/tree/master/samples/react-bot-framework
• Bot Framework Chat React App Customizer | https://github.com/pnp/sp-dev-fx-extensions/tree/master/samples/react-application-botframework-chat
• Learn to Bot | http://bit.ly/LEARN-TO-BOT

Learn more:
• Learn more about the SharePoint developer community at: http://aka.ms/m365pnp
• Visit the SharePoint PnP Developer Community YouTube channel: https://aka.ms/spdev-videos

A

My name's Stephon I'm, technically working at an Austrian based company, sylveon and I'm, mainly responsible for all I call it conversational AI parts within our company. Therefore M nai MVP dealing with about four American things.

A

If any one of you has questions, please feel free to contact me or use the bitly link it down here where we meet Tommy girlist, recon, Russell and happy unshot have made some some videos on that topic, as well as other topics as well to get more into the details. So before we start into actual demos, I want to give you a quick, quick overview of what the conversational AI platform II is actually because it's quite a quite a tremendous platform.

A

Microsoft has built out there, starting with open source frameworks for actually building conversational apps conversational AIFS. With the bot framework. There are a lot of SDKs out there for c-sharp, JavaScript, Python and Java. You can use to actually write codes and build, but on top of that they have what they call or what we call the managed service. So a bot hosting platform in Azure offers you the possibility to actually deploy your bot to the actual resources and connect your bot to the various channels without actually matching the connections between those channels.

A

Yourself, that's done by Microsoft, so you can actually build your bot focus on the the building part development part and then just pick and choose which channels should be offered within your bot solution.

A

So on top of that, what what you actually would need to build a really sophisticated, but our pre trained AI services, meaning cognitive services, offering you intelligence as a service, could be in some kind of vision.

A

Category like computer vision, image, detection or face recognition could be in terms of speech and language, meaning language, understanding, speaker, recognition, speech to text and so on and so forth and those are actually the services which make your bot intelligent, because about itself built with the pod framework, is nothing more than a regular web that you build and then, if you're, quite new to the topic, a lot of templates in solution accelerators out there. So you don't need to start from scratch.

A

You can use the solution, accelerators can download them on the github pages and you pretty much have a Batra up and running within a couple of minutes. We don't need to actually write code, just look under behind the curtains. How Microsoft feels sophisticated thoughts themselves in terms of the boss member itself as we're speaking today of the bot framework composer, which is one part of the Microsoft bot framework. We have a lot of sample repos out there. There are a lot of CLI tools out there for doing automation and stuff. You got your own emulator.

A

You can use to actually build and debug BOTS locally, so without the need to actually building your boat and then deploying it first before you can test it. You can do that locally, and there are a lot of other controls out there for implementing web chat and stuff or in implementing it into your SharePoint pages. Or what have you coming through the implementation side of things? You can build your bot actually once and then connect it to a wide range of gems, meaning you can connect it to teams. You can connect it to email.

A

You can connect it to SMS, meaning text messaging bot, but you can also make use of the third-party channels like Facebook messenger, like slack and so on and so forth, and the cool thing here is really you build the pod once and then deploy it to whichever channel you want to have it deployed without actually needing to build multiple BOTS for the multiple channels, of course, here and there there are differences in the channels, some support.

A

Let's say at that discards, while some don't some support speech, while some don't and so on and so forth, which you need to think about before you actually connect it to the various channels, but in general you could say you build the stuff once you develop it once and then you can run it on most everywhere and but for the topic of today the Microsoft pod framework Composer.

A

This is quite a new tool. You can make use of to build abroad and, as you can see on the slides here, it's basically a graphical user interface for building about. So you don't need to actually write the code yourself.

A

You just use the composer to actually build the bot to define the dialogues design, the bots behavior without the need of writing code, because composer in the backend writes the code for you, and so this is pretty much a tool not only built for developers but also builds for, let's say power, users or business users to actually get get the bot building or what development face.

A

Much quicker and easier, of course, where you can actually start in composer and then hand it off to let's say the development team to make small or sophisticated actions in there or tweak the bot here and there, which you can't do with composer out of the box. So the whole bot development lifecycle gets much more easier with the introduction of composer and composer mainly consists of various pieces. Of course.

A

First of all, the visual editor, where you can just design the BOTS behavior you can design, dialogues and stuff like that, but also what you get with composer is the ability to handle all the things happening around your bot right within that tool as well. So you can manage your language understanding models right within composer.

A

You don't need to actually switch to another portal which, in the past, was the Luas portal to actually scaffold your let's say language understanding model and then go back to visual studio code or visual studio to implement it into your bot. You can do it right within composer.

A

The same is true for the opposite, meaning language generation. This is quite a new topic in the bot side of things where you can actually make use of much more sophisticated. Let the dialogue concepts where you could implement language generation models to make Baltimore make the bot behave more personal, for instance.

A

If, if someone reads the the bot, the bot should read back but depending on the time of the day, the bot will create either with good morning good afternoon or good evening to make the whole conversation much more personal and much more user friendly in a way, instead of just saying hi each and every time the user reads. The bot at a very simple scenario and all of those concepts are basically wrapped up in what what's called adaptive dialogues so meaning in the past.

A

You would need to write your dialogues using either c-sharp or JavaScript or typescript or whatever, and nowadays you with the adaptive dialects concept you just right chase them. So dialogues are nothing more than chasing representatives, which include all of that stuff, meaning language, understanding, parts, language generation, parts and the triggers and actions. You will have in your dialogues, and that makes it much more flexible. So you could basically go ahead, use the composer to just model out your dialogues and then hand them over to the developers for implementing them.

A

No matter if they're using c-sharp today and tomorrow, they're writing a JavaScript, but they can actually make use of you that, if dialogues in either case so the whole bot framework and and what development kind of art architecture is more more holistic in that approach, but enough first light, let's jump into a demo. What I have here is the bot from the composer as I have my slideshow still on. You can't see it, but it's basically a an electron app.

A

You can install on your machine, you can get it from from the heat up repo or, if you want to. You, can also download the source and build the composer tool by yourself. If you want to tweak it, if you want to customize it or if you want to extend it and it within here kind of oriented to what our virtual agent does and kind of of power automate UI, where you can actually make use of that visual editor to design your BOTS behavior.

A

So if you want to build a new bot, of course, you can start from scratch. If you know what are you doing, you can basically get your get your empty butt up and running and scaffold the whole dialogues by yourself, but the cool thing here is: you have a lot of working samples in here, which you can use to actually see what's possible with the composer tool, what you can do with it and what you can, what you can't do with it together, let's say starting point to take it further down the road.

A

So what I actually wanna wanna demo today is a simple but capable of answering questions which are stored in a Q&A maker knowledge base, as well as implementing some language understanding parts within it. So I I'll go ahead with the Q&A maker and Luis template and I'll, give you the name, I, hope I, don't have Chuck Norris already in my composer folder, and there you go right now. Would you get is basically working, but with some some triggers implemented.

A

So, on the left hand side, you have your dialogues and within the dialogues you have your triggers triggers means each and every time the bot detects a language understanding intent. It will trigger that specific part within the within the dialogue, meaning if a user enters the conversation. There's a conversation update activity happening telling the body someone has entered the conversation a member has added to the was added to the conversation, and you would need to do the following: read the user? Send the response welcome?

A

You can type help to learn more there's another trigger for help out-of-the-box. So if a user types in help or I need to help- or please help those kind of intents are basically stored within your language understanding model in each and every time, the language understanding model detects the the help intent. The bot will actually send a response to help the user get going with within the conversation and within the Q&A maker and Luis sample.

A

If the intent is unknown, meaning neither the help intent has been triggered or the by surface help infant has been triggered. It will look up the message from the user in in the Q&A maker knowledgebase see if an answer can be found in there or not, and if, if so, the answer will be sent back to the user and if not no match message will be sent back to the user.

A

So if I'm now gonna go ahead and just update the app settings or the what's called in the bot framework default settings, I need to propose my at least Q&A maker knowledgebase ID. So the bot will actually know where to look up the questions we've. The infant has not has not been detected so I'll go ahead and insert my knowledge base details in there and then, if I'm gonna go ahead and just click start.

A

What the composer will prompt me to implement my Louis primary key, so it can build up the language understanding model for me and after some seconds the bot will be ready to use. So in the background, it will spin up the dotnet core application for me and if the bot has been up and running, I can immediately test the bot right within all right.

A

From my composer tool, I have to test emulator button up here, so what it actually will do is it will bring up the bot, Remick emulator I mentioned earlier, where I can test and debug my bought locally. So if I'm not gonna, go ahead and open the emulator, it will greet me and now I can do things like I, don't know. What's in my knowledge base, but I'm pretty sure, there's something about the bot framework in there.

A

I can go ahead and ask the bot and the bot will look up the question in my knowledgebase not yet because, unfortunately, these seems to be the help in turn triggered so in here, I can see, help intent was triggered and it will bring back the answer from my help trigger, but that's not sophisticated at all. But what we can do in here is define a new trigger and I wanted to define a new trigger for, let's say: I want the bot to tell me a joke. So I have my trigger phrases in here.

A

Tell me a joke. Are you funny? Do you know choke? Please tell a joke and so on and so forth. This goes right into my language understanding model and what I can do right here is I can alter the BOTS behavior or the dialogues behavior, so I can send the response. If you want to hard-code the choke I can ask like a question saying: I wanna have a text. Input number input, confirmation whatever what have you, but what I can also do is send an HTTP request right for me right from within here.

A

So there's this very cool Chuck Norris API, which I can query and I can store the result in a variable, so really easy, HTP request yet just insert the URL in there and just tell the dialogue which property to fill at the end of the day and then I can just at a very simple stage. I can send the response. I get right from within here, so I can refer to dialogue. Choke response is the actual response. I'm I'm getting from the API and the value is within content of value.

A

If I restart the bot now and now, it's building up the language understanding model again, because I've added a new intent in there and I've added a new trigger in there in there.

A

But if it's done build building and publishing I can restart my conversation and say tell me a joke, for instance, and right now it should hit it chuck, norris api and give me back joke and within here we see if it's the api and we get back a joke and therefore we see actually those are basically the most basic scenarios you can do with within box. But they also see you don't need to actually be a developer to start building a.

A

But of course, if, if there are some, let's say really complex dialogues and and stuff like that and you're you're dealing with, let's say conditions or loops, you have to have that development knowledge of course, but you don't need to have the knowledge of writing actual code. You don't need to be a c-sharp or JavaScript developer anymore, to build a bot. If you know the basics and the principles, you can go ahead with composer and just start building your bot right within there. That's it excellent.

B

As an introduction on this and and then we could definitely potentially schedule- and you want a photo later in order to go slightly more details on settings, but this is a great in girona ever started and there was a lot of people kind of a who they were not aware of the whole thing exists, so really really cool stuff. Absolutely so it's always the thing. We're shipping, so many things course to Microsoft. Nowadays, inertia and em 365 science. So it's really hard to stay up to date on what's happening.