Cognigy.AI v4 Update Training

This 1-hour webinar recording aims at users who are already familiar with earlier versions of Cognigy.AI. It covers fundamentals and new capabilities of Cognigy.AI v4 and provides a comprehensive tour of new features and functions. It also shows newly added developer capabilities and provides best practices for transitioning from v3 to v4.

Trainer: Derek Roberti
Playtime: 1 hour
Captions: English, 日本語

Transcript

Sebastian Hello and welcome to today's webinar on Cognigy.AI v4. A training for everyone who's used to work in Cognigy.AI v3 and is now taking a peek at v4. My name is Sebastian Glock and I am Technology Evangelist with Cognigy. And with me is Derek Roberti, VP Technology. Hey, Derek, are you there?

Derek Hello. Yes, hi, everyone.

Sebastian Wonderful, thank you! So everyone in this webinar on mute. We're going to have a dedicated Q&A session by the end of today's presentation. When we're through with the intros, we're going to focus on Derek's hands-on presentation of v4. And then there's the Q&A session where we will answer all the questions that may come up during today's session.

Sebastian So for context, just one brief slide right here. Cognigy.AI, v4. What did we do there? What's behind it? So the entire project of revamping Cognigy.AI was under the umbrella of Advanced Simplicity. So what we intended to do was enabling to handle very complex and demanding conversational AI use cases while making it as simple as possible for business users and also developers to create and orchestrate those experiences. And everything we did falls into one of these four buckets that are listed on the screen right here. So first of all, it's the editing experience. You're going to see in Derek's session in just a minute that editing conversational experiences with the Flow Editor is now more intuitive than ever. It doesn't require technical training or a lot of kind of research and education to create those experiences. It's very easy and simpler to do than before. Developer experience is the second bucket. We're very well aware that not everything that has to do with conversational experiences can be done over a graphical UI or with a low-code interface. Sometimes developers need powerful tools to build the experiences they want, and that includes, for example, a command line interface, a powerful API. And we've done a lot in that direction to make developers' lives easier. The third point right here, conversational experience, that refers to the completely revamped NLU unit that works under the hood. So that is kind of the core AI element of Cognigy.AI. And we were able to increase its KPIs and benchmark significantly in v4 compared to v3. And also the machine learning got a lot faster and is much more performant than you might have experienced that in previous projects. Number four right here, voice experience, that refers to our now built in capabilities to connect Cognigy.AI to phone lines. So the Voice Gateway is a new product that we introduced together with v4 on September 1 that now makes it easier than ever to provide conversational experiences on phone lines. If you want to know more about the context of Cognigy.AI v4, why we did what we did, I recommend you taking a look at our homepage. cognigy.com/cognigy-release-show-recording and there you will see a 20 minutes video of kind of a behind the scenes look what we did and why we did that. And with that I'm very happy to hand this now over to Derek and guide you through today's training of Cognigy.AI v4. Derek, the stage is yours!

Derek Wonderful. Thank you, Sebastian. And can you see my screen OK?

Sebastian I do.

Derek Excellent, so hello, everyone, and I'm excited to share Cognigy v4 with you today. Coming from a Cognigy 3.6 background, you are used to kind of everything, how it works, the way it works, where it lives. And I think Cognigy 3.6, it was an excellent market leading product. So it still thought through very well. Once you spend some time with Cognigy v4 you're going to find that everything clicks just even a little bit better. The way that Flows work, the way the user interface is designed, it really takes the experience of developing conversations to a whole new level, even even more intuitive and more logical than v3.6. And we've made a lot that we did in 3.6 even easier to use. I'm excited to share this with you today. One of the themes Sebastien highlighted was advanced simplicity, and that's the idea of kind of making it easier to do complex things in the product. And as part of that, we've invested in really building in a lot of help into the product itself.

So if you go to cognigy.com and request a free trial, you'll be immediately put into the product. Once you sign up and you'll see something like this. You'll also see a lot of in-product help that's going to walk you through how to do a lot of steps in creating Flows. I want to show you first where some resources live. So this is my new home screen in Cogngiy.AI v4 and we have our news feed just like we did in v3.6. But we also have this "get started" section. This is going to link you directly to some useful content. So as you transition from v3.6 to v4, these can be really valuable tutorials to help you create Flows and update your NLP, create new Virtual Agents within Cognigy v4. We also linked directly to our new Help Center. And our new Help Center. It doesn't replace our documentation, but it's a much more enhanced, application-oriented version of the documentation site.

So we have a few key areas here. One is exploring Cognigy.AI. That's going to be you're getting started guide and how to accomplish some basic tasks within Cognigy. I'm really excited about the how-tos and tutorials. You're going to see a lot of really practical applied scenarios of how to take a lot of Cognigy's new functionality and how to make it useful to particular use cases you're trying to implement in the platform. The community section replaces our old spectrum community. So same idea: You're going to be able to ask questions that might not be urgent support questions, but might be requests for general help or feedback. And our community site will take you there. And then, of course, you can always submit support tickets if you are a customer or partner with Cognigy. So I encourage you to look at that and we will continue to reference that throughout the webinar today. We still have our documentation site at docs.cognigy.com. And you might be wondering, where do I go to do one thing or another? So the documentation site is really there to kind of have more functional documentation. It's very good, complete documentation of how every facet of the product works. But the Help Center, again, is more that use case driven approach to providing information. So I hope you'll find those resources useful and they're easily accessible right within the Cognigy user interface. So to get started, so I've logged into Cognigy. This is that landing page that you're used to in v3.6. And on the left-hand side, I'm going to have what we used to call our projects. In Cognigy v4 we call these "Agents". It's really the same concept in that each Agent describes a set of resources, all of which can see each other. But the reason why we named it an Agent rather than a project is we're really thinking about creating Virtual Agents who are going to be able to handle phone calls for you, who are going to be able to handle chat messages for you. So Agents that are supplementing your team with an automated set of digital workers.

The first thing I want to do is be able to create a new Agent, you'll see I have one here. It's named after my favorite pizza restaurant here in San Francisco. But before I go into it, I just want to go through the wizard that's going to let me build out a new Agent or in v3.6 terms, a new project. You recall in v3.6 we had a simple wizard that let you create a new project. We have more to this now. So I'll set up an Agent with some basic information. I'll call it "training". As before you can choose an NLU language and color.

I'll click next. And now we can choose a template. Like with v3.6 you're pretty much always going to start with this blank template. But for purposes of introducing yourself to Cognigy v4, I would strongly recommend you create a new Agent using the restaurant template. That's what we're going to be looking at today. And it's a really nice way to just illustrate some of the new features that are within the product. But for now, just choose blank. I'll click next. A new thing you can do in this setup wizard, is preselect your channel so you don't have to go in and create the Endpoints after you've created the project. Webchat is selected by default. But you'll see all of the standard channels that you're used to, including our Voice Gateway channel.

Click next again. This screen allows you to add premade skills, and this is stuff that a lot of which was already in v3.6, but in v3.6, they were features that you could enable or disable in a Flow. So we had, for example, pre-built Lexicons for airport codes or countries, for example. Now you're going to select these in the wizard and these will be pre-populated or built-out Lexicons or Flow. The advantage of it is that you have complete transparency into what is in these Lexicons or pre-built skills. So in v3.6, it was a little bit of a black box as to what was actually there. Now you're actually going to see these Lexicons or Flows and be able to modify them yourself.

So once I'm done I can just click the create button, which I won't do in this case because I've already created my Flow. Right here. This is the "Mazzeria" Flow, so this is my pizza Flow based on the restaurant template. And I'm going to spend a lot of time here because it's like I said, a nice way to illustrate how the v3.6 concepts translate into v4. And before I get started, I'll just show you a basic interaction of what the Flow is trying to achieve. I'm opening the Interaction Panel here on the right-hand side.

This is what you're used to and you still just click this bubble icon or dialog icon to be able to open and close the Interaction Panel. We'll go through more of what you're going to see here as our conversation continues. But I'll start off by just starting off the chat saying hello.

It's going to execute my Flow and we'll see a nice welcome message with some happy people toasting each other in a restaurant and welcoming me to Cognigy eatery. And I'll also get some options that says "Welcome to Cognigy Eatery. How may I help you today?" And it's kind of let me order make a reservation or a few locations. So this is going to look familiar. One thing I want to highlight is these orange bubbles.

These are new. And I'll talk about what they mean in a second or in a few minutes. I am going to go ahead and make a reservation. So I'll just click, make a reservation. I'm going to be visiting the San Francisco location. I can select my date and time. Say "how many people will be attending?" And then enter my email to get a confirmation message.

And great, now I've got my reservation all set, so that's kind of the basic interaction that we'll be looking at today within our Flow. So let's talk about the new UI. Let's talk about what you're seeing in front of you right now. The first place I want to get started is how to find some of the stuff that you're used to. So what you're going to do is this left-hand bar is going to be something where you're going to spend some time finding your resources. So we've categorized a lot of the Cognigy resources into different groups. Build is going to be your most common one, and this is where you're going to access your Flows. So you'll see I have three Flows listed here.

I also have my Lexicons, my connections and my NLU connectors. So conceptually simple, but this is just where you're going to go to find these common resources for Endpoints, if you go under deploy, you'll see we will have your Endpoints defined here and you'll see that I have the Webchat Endpoints it up. That's going to look familiar to you. We also have a new capability called Snapshots, which we'll talk about later. Under Test we have Playbooks, which you're familiar with, and now we make it just a little bit easier to find where the logs are. Sometimes that wasn't easy to see in Cognigy v3.6. And then under Tweak, we have the Intent Trainer. Analytics are straightforward that's going to take you to our analytics reports. Under Manage there's a couple of things that are useful to note. One is you have your settings that's familiar to you from v3.6. The other thing is this is where your Contact Profiles live now. They used to live under this icon and the top right corner, but now they're going to live under Manage here.

So let's go into a Flow and talk about where to find some other important things that you're going to be using. So we have the kind of the Flow Editor divided into a few different kinds of conceptual interfaces. So on this, we have the Flow Editor itself, which is self-explanatory. You access this through this chart section at the top here. The next category is going to be the NLU. And this is where our Intent Editor lives. And we'll go into much more detail about what's new and the Intent Editor, in a few minutes. And then you have your settings. So again, and thinking about where your eyes are going to go and where you're going to click. This is going to be a place where you're going to navigate most commonly between the chart view and the NLU view.

The user interface itself has a lot of nice new navigation elements to it. One of the most obvious changes is that we've gone from a kind of left-to-right orientation for the Flow to a top-to-bottom orientation of the Flow. And we did this because we felt like this based on feedback we got from users and usability experts that that's kind of a little bit more how people's minds work. Since you are used to v3.6, it'll take a very short amount of time, and then you'll get used to this orientation of the Flow.

Some really neat features we have built-in is the ability to pinch and zoom, so if your Flows get really big, you can zoom in and out of those really easily. And you can always, of course, in the bottom left corner use your zoom tools and your centering tools here as well. Really neat thing is the ability to control-click on any Node and that's going to expand this menu of letting me do things like copy-paste, cut, delete and collapse. This is a new feature: so here I have this Lookup Node, we'll talk a little bit about what that means, but if I click collapse or just double click on the Node itself, it's going to compact that whole part of the tree to make it easier to kind of view and understand my Flow. Some other things that we can do is we can copy-paste branches, so, for example, if I copied this Lookup branch and pasted it over here, for example, it's going to paste that entire branch in all of its children. To add new Nodes we'll click the plus sign and I'll talk about adding new Nodes in just a few minutes. To edit a Node in v4, you click on it, so I'll start off with this title card, and a welcome message, and you'll see this little pop-out view. I'll talk more about how Say Nodes work, which is what this is in a few minutes. But just for now, from a user interface perspective, if you're going to click on it, it's going to open this view here.

Now, new to Cognigy is this idea, this is the compact mode for quick editing. You can also go to the expanded mode. So if I click this little arrow, it's going to pop out to this broader view just so I can have a bigger workspace and see what I'm doing. In the case of these cards, I can add more cards by clicking here. So having this expanded interface is kind of handy to be able to do that. So you can scroll back and forth in the collapsed interface, you'll see I can still see all those different cards, but they're just a little bit of a collapsed view there.

For adding Nodes, if we've revamped this in a way that I really like, so in v3.6 you would click on the three dots to create a new Node and you'd have these hierarchical categories. We still have that, but I think we've dramatically improved the experience. So if you click on a plus sign to add a new Node, you'll first see a list of the top Nodes that you're going to use, a Say Node, a Question Node, an If Node, and a Lookup Node. So instead of having to go to the different categories for these things, boom, they're right there at the top for you to be able to access. We still do have the categories. So you have a Node from the logic category or if you want to update profiles, you can navigate the categories here or you can do a search. And honestly, this is what I do most of the time, even for Nodes. So I could just click on just because this is how I work. So if I want to go to Lookup Node, that's going to quickly take me there. If I want to find a Say Node, it's going to quickly take me there. A Question Node is going to give me my question options there. So lots of different ways that you can access the different Node types within Cognigy, which I think is much easier to use. We also have Extensions here and we won't talk a lot about this today. But the Cognigy Integration Framework has been replaced by Cognigy Extensions, and we've made it even easier to be able to upload those integrations with external applications within Cognigy and be able to add them to the tree just using this user interface.

All right, so that's kind of a basic walk-through of the UI itself, what I want to do now is start drilling into some of the new Node types and concepts. So similar to what you're used to in Cognigy v3.6 is we have what we used to call a switch Node and I don't know about for you, but because we had a Switch Node and a switch Flow Node, so many times I would click one or the other and really mean the opposite. So we've changed the language a little bit to make it avoid that and make things a little bit more intuitive. So the old switch Flow or Switch Node is called Lookup. Now, there's a different word for it. When you create a new Lookup Node by default, it's going to select Intent. And you can, of course, use this dropdown menu to select the other intense like you could in Cognigy v3.6. Another kind of Node that we have is a question note, and this is a really big change within Cognigy. So I am going to switch actually over to our table reservation Flow, so I'll give it back to my Flows and go to table reservation. And you'll recall in the demo, I said I wanted to make a reservation, it was going to be in San Francisco on this particular date with this many attendees. Now, we did this within a table reservation Flow. This is going to look a lot like what a process looked like in Cognigy v3.6. What we've done in Cognigy is we really merged the Flow concept and the process concept into one user interface. So what you're looking at here is actually a Flow, just like you're used to in Cognigy on v3.6. But we now have questioned Nodes that are going to walk people step by step through a process. We had this to a degree in the Flow Editor in Cognigy v3.6 and that we had Slot Questions and yes, no questions. But the new Question Nodes support a lot of different question types, so I'll click on this Question Node here.

This is a Slot Question, but if you choose this dropdown, you'll see a lot of different options for these question types. You can, unlike in processes, you can now use the full capabilities of Say Nodes, you can do quick replies or galleries or do channel-specific output here that you couldn't do in processes and v3.6. And there's a number of configuration options here, but to just go over the logic of how it works is, I'm going to start this Flow. It's going to ask me a question. And like with processes and v3.6, it's going to wait for an answer. And that's going to ask me the next question. So instead of starting the Flow execution from the beginning. That's going to start it from the last question that was asked.

There's, of course, Say Nodes, which I've shown you already, the other one I want to show you as a renamed Node type is going to be Goto. So you'll recall in Cognigy v3.6, we had Switch Flow and Execute Flow, and this was a way to be able to move usually from one Flow to another, which allow us to create modular Flows and simplify our Flows. So instead of Switch and Execute Flow, we now have a Goto. So in the demo I showed, I said I wanted to make a reservation that would trigger this reservation Intent. And now I want to take the conversation over to the table reservation Flow, so I'm going to click on this, Goto Node. And here's a really, really neat change, which is now I can choose to go to a Flow, but instead of having to start at the start Node, which is how we did it in Cognigy v3.6, I can choose any question or other Entry Point. We'll talk about Entry Points in a bit. A Question or Entry Point Node within that Flow.

So it's a really powerful way to be able to give you more flexibility in how you route users from one Flow to another and what point in the conversation and those other Flows you want them to start. You have the advanced options of this execution mode, which is going to let you either go to this target Node and continue the conversation or go to the target Node and wait for input. So that's kind of the Switch Flow and Execute Flow concept. We have the v3.6. So some nice new features there.

A fine point that is interesting to note is that when you do click on to Goto, and I do this dropdown and it shows me question email, question location. It's taking this value from the label that you create in the Node. So if you don't do this, you would just see it's a question, question, question. But if you create labels on your Node, that's going to show those labels here. So it makes it really easy to know which note you're navigating to.

OK, so building on this idea is that, in Cognigy v3.6, we would have inputs in your normal Flows, not in processes that would be process from the start Node onwards every time. So it's like an input into Cognigy It goes from the start to the end. The next input we can go from the Start Node to the end. The exception to this was the yes-no questions and Slot Questions and also processes. But as I mentioned, Cognigy merges all of these features into the single interface.

To kind of highlight this visually, you'll see like in the start Node, for example, this little green arrow marker. What this says is: this is an Entry Point. So this is a place where inputs can start in this Flow. I'm going to go down to this Question Node here, it's an optional question, and it also has a little caret. This one is gray, so it might blend into the background a little bit. But this is going to be another Entry Point. So conversations can start or inputs can start anywhere that you see one of these Entry Point icons.

All right, let's look, I talked a little bit about some new names for Nodes. One is we saw here that a switch Node is now called a Lookup Node. I mentioned a Goto Node, replaces the switch Flow and execute Flow Node, Question Nodes, essentially replace processes that we had in Cognigy v3.6. Some other small changes here is if you go and find stop. Let's see here.

Well, I'm in a particular place. Here we go. I go to a stop and return note, this is the same thing as the old Stop Flow. And Sleep is the same thing as the old Pause. They're just different names to make a little bit more intuitive to use. Now, Say Nodes, as you know, is how we output something to the user and we've added some really new powerful behavior to how Say Nodes work, I showed you this already, but there's some really interesting nuances here.

So first off, in creating these visual elements here, the first thing you're going to do, let's say, create a new Say Node, is you're going to choose the output type. Do you want it to be text, text with quick replies, galleries, text with buttons, so on and so forth. And in this case, I chose a gallery. So let's look at how to edit galleries. In this compact view, if I just click on the image, I can put a URL for the image, I can type in my text here. If I want to add additional buttons, I can add them here. And in my expanded view I showed earlier, I can add new cards as many as I want to.

So it's a really nice visual editor for being able to create these visual elements, it's a little bit more user friendly than we had in v3.6. But one thing you might see as missing is, well, how do I output this for the Webchat versus Facebook Messenger, for example, and in Cognigy v3.6, you would do that on separate channels. In Cognigy V4.0, we've added intelligence here, where when this Say Node is executed. Cognigy is going to look at what you've created here, I've created this gallery widget and it's going to put it into the right format for the supported channel.

So if I have a channel that's going to support this kind of output, regardless of what that channel is, Cognigy will automatically create it and translate it into the format of that channel. Now, if the channel doesn't support a visual component, let's say it's SMS or let's say it's a voice channel, we have the option to put fallback text here. And in the demo, I am just going to show you again: When I said "hello", we saw the visual elements, but we also saw this orange bubble. That's your fallback message. So that's what a user is going to see if they're on a channel that doesn't support this text. Another nice one to look at here in terms of the Say Nodes is going to be text and quick replies, this has changed significantly. It's easier to use, but you just need to understand how to use it. So if in my Say Node, I select text and quick replies. I click "add a reply" and what you'll see is that adds a button here that says empty reply. When I click on empty reply, it's going to give me access to all of these fields, such as the button title and the postback value. So these are the same fields you saw on v3.6, but it's just a different user interface, a little bit more visual in terms of how you create these quick replies.

There are some cases where, you know, it's really nice that Cognigy is going to try to figure out if a channel supports this and translates this into the right channel output. But there's going to be some cases where that just doesn't make sense. So, for example, if you're using Twilio or Audiocodes and you want to output some SSML or custom JSON values, that's just not going to really work within this concept of translating this into a format of the target Endpoint.

So what I can do on the channels is I can click the plus sign and for example, I could add the voice gateway channel and for my voice Gateway Channel, I can choose text, I can choose custom JSON and I have an editor for it here. So in Cognigy v3.6, the philosophy for channels was create your output for each channel that you're using, in Cognigy V4.0 it's really create everything in this AI channel and then for the exceptions, things that are going to be different, then you can set up unique channels such as in this case, the Voice Gateway Channel.

All right. So this is going to be some of the basic new Node types, I'm going to show you some more. One more thing in the user interface that's going to be helpful to see is appear in the standard set of buttons. We have we have a new button called the Task Monitor, and this is going to show you what jobs are in progress. Most often this is going to be training the NLU, but it could also be uploading Lexicons or Intents or Snapshots. You'll be able to see that feedback right here within the tool. Some important Node types I mentioned earlier are Question Nodes. I'm going to add a new Question Node here, just to give you an example of how this works and how to handle the responses.

So. I'll just kind of choose a part of the tree here, which is this first time execution is actually worth mentioning, one other Node type that's changed is the Once Node. So we still have the Once Node, just like we did in v3.6, but the UI is just a little bit clearer. So I'll say "on once" on the first time, the first execution, we're going to go down this path and the tree and subsequent executions, we'll go to this afterwards. But just for illustration, I will say on the first time I'm going to create a new Question Node. So I click my plus sign here, I can just go ahead and quick question or optional question also click it here. It's going to give me this Node type and I will open it up and this is going to be a yes, no question, I'll say "are you hungry?"

Like that, so just a simple question it's going to ask. You'll recall in v3.6, we would automatically populate and on yes or no or in Slot Questions we would do on answer. Because the Question Node so much more rich now, we don't do those default types. You can change a Question Node, for example, from a yes / no question to a Slot Question to another kind of question. And so having those default child items just doesn't make sense. But we still need to be able to handle the response. So it's going to ask a question and you'll see you might be able to see this great little arrow that's going to say this is an Entry Point now. To respond to the answer, I'm going to add an If Node. And we've added this token, which is really helpful, so if I click on my If Node, if it's going to work the same way that you're used to in v3.6.

So with this. There's a new token called Last Question Result, so I'm going to say if the Last Question Result is true, that means they said "yes", then I will go down the Then branch of the tree, and if it's not true, I'll go down the Else branch of the tree. If I open this up, if you look at the token itself, if you hover over it, it's just a result field and the Cognigy input. So it's nothing too fancy there, but it's just a nice token that's going to let you be able to come up with your responses to these Question Nodes. If you have some conditional branching that's going to follow from them.

The one other thing I will mention is that we have two different Question Node types, I'm going to delete this If here, which is we have questions and then we have optional questions. And an optional question is, as it sounds, which is a question. It's kind of like in the processes we had a similar concept as whether a question was required or not, I believe. An optional Question Node gives me some additional features here, so if I go to Advanced, I can say what is the "forget question" threshold. And then that's going to let the Cognigy kind of keep in mind this question and wait for user input, but it's not going to require an answer. Whereas with the Question Node, that's going to require that you respond with one of the expected response types.

So I am going to clean up my Flow here a little bit and just doing my control click to delete these Nodes. And we're back to where we were. A brand new feature in Cognigy is something called Slot Fillers. So I'll kind of refer to the documentation, because it's going to be probably the easiest way to walk through this when you have time on your own. But in Cognigy v3.6, we introduced something called the missing pattern. Some of you may not have ever used the missing pattern, but some of you have the concept of the missing pattern is this in normal human interactions, we sometimes convey information all at once and other times we convey information one piece at a time. So, for example, I might say I want to travel with three people to San Francisco next week. So in Cognigy terms, wanting to travel might be your Intent or wanting to schedule a trip and then you might have Slots, which is three people might have a location, thought that San Francisco and next week would be a date Slot. So I kind of give all that information at once, or I might say it one at a time. So I might say, hey, I want to travel. And the boss might say where to? I will say to San Francisco. And then it would say, how many people? And I would say three and so on and so forth. So to support both those kind of linear and all at once scenarios, we had a concept in Cognigy v3.6 called the missing pattern and the pattern involved creating a basic Flow. That was that was kind of easy to understand, but it also required some code notes which were less user friendly. So we wanted to be able to bring in that same capability and 4.0, but not make you write any code. So we've done that using what we call Slot Fillers and what Slot Fillers do is they. Keep track of Slot matches that are coming in throughout the conversations and at stores, those in the Cognigy context, so normally Slot matches are just in the Cognigy input, but if you set up Slot Fillers, it's going to store those in the Cognigy context. And when we use those in combination with question notes, we're able to ask questions or skip questions based on what we see in the context object.

So if this sounds like a scenario you need to tackle, I encourage you just go to the Help Center and go under tutorials and best practices and learn about how the missing pattern works with Slot Fillers and Cognigy v4.

In the user interface, Slot Fillers are going to live under this NLU section and you go to Slot Fillers here and then this is where you set up your Slot Fillers. But since we're here in the NLU section, let's kind of walk through the new Intent Editor. It looks the same, but it's a lot different, and I think the user interface has improved and that's a lot more powerful. So I'm going to go to my reservation Intent. And this is what I was interacting with in the original demo. I said I wanted to make a reservation. I have my example sentences here. So this is going to look very familiar to you already. But here's something interesting that we've added. It's the ability to have annotations. So let's imagine that a user might say "do you have locations in Sydney?"

What I can do now is I can highlight Sydney, because that's a place name.

I'm going to click this plus sign over here on the right, and it gives me this new dropdown. And what I can say is Sydney is a Slot match and it's going to be a custom Lexicons Slot that I've created. And it will be a location. This is a really nice way to annotate this example sentence, to be able to say, if they say, do you have locations in San Francisco or Los Angeles or wherever, it's going to treat this example sentence as something that's kind of like a sentence with a placeholder, which is Sydney for a place name. So we had a similar feature, pretty much the same feature in Cognigy before, but you added these double brackets to be able to add Slot matches. So you might you might remember this, but the user interface is just much simpler to use and much more intuitive, I think, in Cognigy v4. We've also introduced and I'm going to back out of this and don't think the changes there. We've also introduced what we call Instant Feedback in V4 that's incredibly powerful and unique feature. Once you create your Intents and train them, and then every time you make changes and train them, Cognigy is going to give you these red, green and yellow lights that are going to show up within the user interface. And these show up on example sentences, they show up on intents and they show up on the overall model itself. This is a really nice way to be able to get more of a holistic view of what is the quality of my example sentences? Am I on the right track? Are there overlaps? Am I using a sentence and one Intent that's really similar to a sentence I'm using and another Intent. So this is a nice way to kind of give you a sense as an Intent developer to see how was the quality of my Intents. You're not necessarily aiming for everything to be green. You can have yellow Intents and you can even have red ones as long as they're kind of meeting the benchmarks that you want for your Intent accuracy. But it does give you a nice feedback in a way to zero in on places where you might want to do some work or improvements. You can find more information on how to understand this feedback in the documentation under Intent Analyzer. And this will kind of give you more detailed information about how to interpret those red, yellow and green lights.

Another nice feature we've introduced in Cognigy 4.0 is Hierarchical Intents. This may not have been something that you used in Cognigy v3.6, but if you had a large number of intents and v3.6, you could improve Intent detection through a multi-step approach. So in your starting Flow, you could group Intents into categories that would have all of the example sentences that might be relevant to a category. And then when the category Intent was matched, you could use a Execute Flow or Switch Flow to route the user to a secondary Flow where that detailed Intents will live. This required creating multiple Flows and having some duplication of example sentences across different Flows so it optimized the Intent detection, but it required a little bit more work. In Cognigy v4 we've eliminated the need to handle this multistep Intent detection and Flows, and we now offer what we call Hierarchical Intents. So I'll give you an example of this. To create a Hierarchical Intent, you're going to first create a parent Intent. In this case I'll create one called menu ingredients. Next thing I'm going to do is I'm going to create a child Intent.To create a child Intent here, and let's imagine that the people might ask, do I have a vegetarian options at the restaurant? I might create another child Intent that says, are there gluten free options? And another child Intent that could be asking about low fat options. So imagine that I've created example sentences for all of these, like you would normally do with a standard Intent. On my parent Intent, there's a Switch here that says "inherit example sentences from child intents". And what that's going to mean is it's going to enable this multi-step Intent detection. So in the first pass, the NLU is going to figure out, do I have a close match to one of these child Intents? Once it sees that this is a good category match, then it's going to do this detail level match within the category. So it's a way of refining the NLU. All of this happens in real time, obviously, and it gives you a really nice visual way to organize these. So instead of organizing these into different Flows, you can really just organize these into the hierarchies. So really nice feature.

Last thing I will mention about the new capabilities is the upload / download capability still exists. If I just click these three dots, I can upload Intents and download Intents. The format for upload and download has changed just a little bit. So my recommendation to you is create some Intens like we have here. Just download the Intents and you'll see the new CSV format or you can just go to the documentation site. You'll see it there as well. So you have a lot of fun playing with the new NLU. I hope you enjoy it.

All right, let's switch gears and I want to talk a little bit about versioning in Cognigy v3.6. So we had a really nice versioning feature in 3.6 Where every Flow could have a unique version. And this covered some scenarios well, but it covered other scenarios not so well. So a few things. One is in Cognigy v3.6, I could only version Flows, I could not version processes, for example, and I could not version Lexicons, for example. So there's a limit to what I could version and v3.6. The other thing is when you have multiple Flows in v3.6 and you might be switching from one Flow to another, there was no way to kind of version a combination of Flows. So, for example, if I want version 3 of Flow A and version 10 of Flow B, there was no way to kind of version that combination of Flows. So for that reason and others, we introduce the idea of Snapshots. And what a Snapshot does is that versions everything in the project. So it's literally like its name sounds. We're going to take a picture of all your Flows, all your Lexicons, all your Connectors, everything that you've got into a project, into a Snapshot that you can move then from one environment to another. So I'll create a Snapshot real quickly here. I'll just create a new Snapshot. I will name this "training" and maybe give it a date, 2020-11-05. I'm going to add a description if I want and I'll save this and it's going to create the Snapshot. You'll see, I mentioned this task monitor icon up here. It's going to tell you that it's creating the Snapshot right now. And that's all I have to do to version this project. You will see once the Snapshot is created. There you go. It's done. In my interaction panel, just like I could with versions, I could choose a Snapshot to execute my inputs through. So it's very similar. But the interesting thing is, again, the Snapshot, it's not just a version for a Flow, it's a version for the whole project. So I'm good there. If I want to restore this, so let's say I moved this Snapshot from one environment to another. Or if I want to move it from one environment to another, let's take that step first. I can download the Snapshots. So if I click on the three dots here, you'll see I have this prepared download option. It's going to click, prepare, download. So this is going to trigger Cognigy to do some work behind the scenes to get the Snapshot in a format that it can be exported in.

And then once this is done and we can see our task manager working here to package up the Snapshot. We'll be presented with another option to download the Snapshot, and that's how we move it to another environment. Just give that one more minute.

Once I download the Snapshot, I can move it to another environment or move it to another Agent, similar to what you would think of Cognigy v3.6 and I can upload that directly using the same tool. Once a Snapshot is created, you'll see an icon up here that's kind of let you do some upload capability so you can upload this into the environment. Now, an important exception to note on everything I just said about Snapshots is Endpoints are not added to the Snapshot. And this is a good thing for two reasons. One is you don't want to restore a Snapshot and have it wipe out your existing Endpoints. So that's the reason for it. The other just kind of functional reason for it is all of your Endpoints have unique IDs that are associated with them that allow Cognigy to handle inputs and know which channel and Flow to point them to. So we can't have kind of a duplication of those same ideas within the same system. So those are Snapshots,.

Endpoints. I won't spend much time on creating an Endpoint because it's going to look the same as what you're used to, I just create New Endpoint and I have my drop down menu that lets me choose a specific Endpoint type.

The Intent Trainer has been improved in some interesting ways. I think one thing I really like the most about the Intent Trainer is this: In the old tool, I would have all of my unhandled Intents. And I could add this Intent to the NLU for a specific Flow. But what we've introduced here is if I click "Add to Intent" now, I can actually modify this text before I assign its to an Intent in a target Flow. So if I want to assign this to the locations Intent, for example, I might want to simplify this based on some standards that I'm following for my Intent. So if I don't want to have "are you" on there I can just delete this, maybe delete my punctuation if I'm not using punctuation in my example sentences and it saves this. This so just makes this more powerful and lets you tweak things before you can actually assign them to a Flow.

Couple more features I want to show as we are running out of time here is localization. This is an incredibly powerful new feature. In Cognigy v3.6 you could do different localized versions of Flows, but it would really require copying Flows or using some other tricks to be able to have a multilingual Flow. What we've added in Cognigy v4 is the ability to create new locales and to translate your Flows. So here I have a English locale and a Spanish locale and I could create any number of new locales that I want and you'll see all of our default NLU languages here. Once I create this locale, I can go back to my Flow. I'll go back to my Main Flow here. And at the top, one thing I didn't mention is this dropdown for locale. So I'm looking at the EN-US version of this Flow. I want to go to the Spanish version of this Flow. And what you'll see is my Flow Nodes are grayed out and the ones that are grayed out have not been translated. So if I want to go, for example, to my title card. Here I can add a localization, so, for example, if I want to copy this from the US-one as a starting point, I can do that. And now I can edit this and edit this in Spanish so that we will have those Spanish language output. So it really realizes the vision of being able to create a Flow once and be able to translate that into any language. Or you might have different locales or are maybe just different regions within your organization that might need slightly different outputs. And locales are a great way to handle that.

The final thing I'll cover before we go to the Q&A is just to mention the Extensions as a capability within Cognigy. We used to have the Cognigy Integration Framework. We have renamed this, but for good reason. Cognigy Integration Framework gave you the ability to create custom Nodes in the Flow Editor. What we've done with Extensions is we've really built-in the idea that you are literally creating Flow Nodes in the exact same way that Cognigy creates Flow Nodes.

So the framework for for what we do and what you do has been entirely synched up. So you have all the capabilities that we have already within the out-of-the-box Flow Nodes available to you.

You'll see documentation here on how Extensions work, how to create them, what out-of-the-box Extensions we already have. And if you go to our GitHub site, if you've worked with Extensions before, you'll see they are listed not under custom modules, which is where the Integration Frameworks were, but under Extensions. This is going to give you really detailed documentation and some really nice examples that give you starting points around how to create different kinds of Extensions. But you'll see this is our current list of the out-of-the-box Extensions that we provide.

Two quick points. One is in our GitHub, you will also now see something new: The Cognigy CLI. this gives you a command line interface. If you want to do anything within Cognigy just through the command line, this gives you some tooling to be able to do that. Full documentation and downloads are available here. What makes the CLI possible is that we've radically opened up our APIs to let you do literally anything in Cognigy now with the API. If you go to our API documentation, it has a new home. You can find it by going to the doc site and you'll see everything you can do now. Authentication, administration, analytics. You can manage Flows themselves and individual Nodes within the API. You can manage Endpoints within the API. Lots that you can do here that you could not do before with Cognigy v3.6.

So that's a quick overview of a lot of the new features that are here within Cognigy v4. I hope that you will be able to take some time to dive into some of these themselves and be able to learn more about this really impressive and easy-to-use new interface that realizes the vision of Advanced Simplicity.

Sebastian Thank you very much, Derek. That was a pretty impressive session. And just on time, so thanks for that. Just before we finish this with the incoming questions, just again, as a note, you can try out Cognigy.AI v4 yourself. We have the free trial available on our website. You find it in the top right corner. It is instantly accessible. Just leave your email and your credentials. And you're going to be able to play around with the exact same features that Derek has just presented in the v4 training.

So with that, I have currently three questions coming in and I'm going to go through them one by one. If you have any more questions, please don't hesitate to put them in the chat right now. I'm sure we'll find the time for question number four or five.

So first question, Derek: Is Cognigy.AI v4 only available in a cloud environment?

Derek Just like with Cognigy v3.6, you can have this in Cognigy SaaS platform, our managed hosting platform, or hosted in your own environment. What I recommend just for learning purposes, if your company hasn't upgraded to Cognigy v4 yet, is you can set up a trial account like Sebastian mention, get started immediately and use that as your learning environment as you get up to speed on Cognigy.

Sebastian Great, wonderful. So second question: Is there an API available for handling the Snapshots? I think referring to the idea of automating DQP environments and automating maybe security copies of entire Virtual Agents. So is there an API for Snapshots?

Derek There is. It's very exciting. You can see it in our open API viewer that's linked to on the documentation site. And I actually would love to show that maybe if you ask me another question, I want to show that real quick, if I may.

Sebastian Sure.

Derek And actually, I found it here real quick. Quick way to do it is, you'll remember in the v3.6 documentation up here, we had some links to the guides and the APA reference here. If you just click down on API reference, it's going to take you to this page and it's going to show you kind of where to find the 4.X and 3.X. If you click on the 4.X, it's going to take you to this API viewer. And here's your Snapshot API that you have available right here.

Sebastian Great, wonderful. So final question, I might remember that you referred to that briefly. Where in the toolbar are the processes?

Derek There are no processes. So the processes have been replaced in Cognigy with Question Nodes: so you can add any number of Question Nodes and it's all integrated within the Flow editor. So you can do everything you could do in processes. But what you have now, because it's in the Flow Editor, is you can do channel specific output, you can do galeries, you can kind of do more rich outputs than you could in processes themselves. Having said that, based on your question, that might not make sense now. But as soon as you start using the tool and adding Question Nodes, it's going to click really quickly on how we've integrated those two.

Sebastian Right, that makes sense. OK, let me switch back to the slide real quick with our email addresses. So with that, thank you very much, Derek. Thank you for this impressive session again and also "thank you" for the audience. I was very pleased to see so many people coming to our training session today and feel free to contact us. If you have any questions, we're going to be able to answer those. Don't forget our community as well, which is part of the Help Center. So if you do a posting in there, we monitor that very closely and we might be able to help you in the forum as well, and at the same time help other users who can see the information that we provide to you in our public forum. So thank you very much again for for for having us tonight. Have a great day. I'll stay healthy and talk to you soon. Bye bye.

Derek Thanks, everyone.


Comments

0 comments

Please sign in to leave a comment.

Was this article helpful?
0 out of 0 found this helpful