Virtual Agent Academy: Paris Release Preview
hi everyone good morning good afternoon good evening welcome to virtual agent academy um we are going to get started today and talk about paris release readiness and i know you guys are going to have a lot of questions so i would encourage you to post them into the chat we will be taking pause breaks to review any questions you have as we go through some of the new features today so just a reminder we have a new idea portal out on the community site and this is where if you identify you know enhancements or requirements that you find are not being addressed in the product i would encourage you to post these requirements and enhancements out to the ideas portal that way customers can go out on it they can vote on it they can add commentary it's available on the community and it's one of the menu items so just look for idea portal on the community and feel free to share with us out there so just a quick statement on safe harbor since this is the paris release we are under early availability as of now so a high level feature summary and the release features really fall into four distinct buckets so first we'll address some personalization branding capabilities and how this enhances the user experience as they connect to a live agent we will review some virtual agent designer enhancements as well as some analytics we'll take a look at some of the enhancements around natural language understanding as well as language support and then finally apps that will be available on the store so this is new for us with the paris release we will be um beginning to release some of the features out in the stores apps and those will be available in q3 and q4 of 2020. so first covering user experience and some of the new chat features this is one of my favorite areas because again branding and personalization really enhances the chatbot experience and a lot of organizations do find value from out-of-box branding capabilities so we've listened we've heard you and with the paris release what you can expect is um the capability to to really extend the profile branding for the bot um out to multiple different portals or sites so you might want to have a different bot profile for the hr portal versus the it portal so that's all possible with the paris release um on the list of virtual agent topics where you see the show me everything menu we are releasing a search bar so users can select topics but they can also type in search so they can look up terms they can you know use the search bar to limit the list of topics if it's a large list of topics so that's again new for paris um this is a smaller one but a custom logo in the browser tab so the ability to be able to see the logo um you know and view that in the browser so just checking sounds like the chats are being answered and they're coming in fast so that's great um additionally configurable chat menu options so again you know just more options in terms of how you want to display the chat menu um and things like proactive chat notifications in the virtual agent client so you can now include a notification that lets the user know hey you have an approval due or um you know open enrollment is around the corner or we're switching um you know from from zoom to microsoft teams right so you can notify the person of updates to make it a little bit of a more proactive experience with that we've improved the user experience in microsoft teams so we've added things like estimated wait time to connect with a live agent and under limited access for paris is the voice to support virtual agent again this is for ios it's under limited access but we we do plan on you know looking at releasing that at a future time so supporting live agent chat and this is more for the fulfiller role um agent whisper chat is really a peer-to-peer chat feature so the use case would be i'm helping somebody with a benefits question that comes in and i realize i don't know the answer and i want to reach out to a friend or i want to um you know ask my manager can you help me with with this question so i can get back to this person so that's really what we term whisper chat and so the ability to then download the transcript with the requester as well as between the agents or the agent and the agent manager is available with agent whisper chat again that does require the use of agent workspace or hr agent workspace to use this also under limited access is the dynamic translation for agent chat support so you've probably heard of this or seen this in different parts of the platform specifically around knowledge and this is under limited access for it customers to use for agent chat purposes so it would apply the same dynamic translation functionality um to to agents connecting with a user the user is speaking french the agent speaks german and they can have a conversation in their respective languages so moving on to the designer view so the virtual agent designer the analytics this is one of the more technical sessions so i would say um you know as we look at the designer we've added some builder capabilities here to enhance the virtual agent conversation experience so one of the the exciting areas we've added is a descriptive carousel builder so you've probably seen the carousel before where you can scroll to the right and choose um you know a laptop or a phone or you know an it device of sorts so we've added the ability to include descriptions there so the user will know oh hey this is the one with the larger memory or this is the one that has um you know this particular adapter i've been looking for so the descriptive feature helps the user make a choice without having to build that into the virtual agent flow we've also added a file picker user input control so again we're expanding the ability to upload images to now pdf documents excel documents word documents um so if somebody needs to for example they're requesting a leave of absence and they need to upload um you know a document from a like a doctor's note or they need to upload any kind of paperwork virtual agent supports the upload of the actual file as opposed to just an image so we've expanded the abilities around that we also have what we call custom input and response controls so this is really you know up to your creative imagination um we allow you to build it into a topic block but it's really a way to say do we want video content do we want you know something super unique to our organization that we can include as part of virtual agent so you know it really would require more of a developer skill set to architect into deploy but we do give you the capabilities to um you know get creative with the type of of inputs and response controls that you can use what's really exciting is the text format validation um this extends to you know if on the user input nodes if somebody types in an email address or a phone number oftentimes the input may not match the exact syntax or the taxonomy of what that that input needs to be so for example if i need to update a personal email address and i type in marcel at you know servicenow.com with 2ms that is an improper email validation so you know what we want to do is is give you the ability to add email validation phone validation so that the user is required to enter a specific format so that when the update is made the data is accurate and the integrity is preserved based on that validation feature for it is a concept of virtual agent lite so these would be you know a subset of topics available for all customers to explore to test out the capabilities of virtual agent um it's mainly geared for those not using virtual agent today to kind of see you know are we seeing a lot of value from this are we excited about it and you know do we want to begin using virtual agent for our organization on the analytics side we have some additional tabs included with the dashboard and those are designed to build out on the original dashboard that we have today so this will include you know data around interactions with conversations what channel is the user accessing virtual agent from what type of topics are they requesting and then anything around user interactions um as well as bringing in nlu data to that dashboard so those are some additional enhancements we've made on the dashboard so skipping ahead to natural language understanding and this is a feature some customers turned on today this is a feature some have not um but to mention some of the enhancements with the paris release what we've done is we've really enhanced the model builder so you'll see a lot of terms like synonyms vocabulary entities really what we did is we we made the model builder a little bit more intuitive so um previously if you wanted to add a vocabulary to a base word right so your base word could be um application and maybe as part of vocabulary you would have microsoft teams you would have zoom you would have slack etc but you were capped at a limit of five so we've removed the limit on the on the um vocabulary so you can expand the the list of applications or the list of um vocab terms you want to include we also have a concept of an open open-ended entity so this just lets you um use entities you know in in a broader sense than previously used they were you know previously a simple entity a list entity a pattern entity so now we have a concept of an open-ended entity that can be applied in different ways we've removed the case sensitivity of vocabulary so previously you might have had utterances that had capital words they've had different variations of how people would type things and so we've removed the need to add you know case context and capital letters and to the vocab and the synonyms so additionally the table lookup type has been added to the synonyms so you can actually source from a table now so going back to that use case of application if i have a list of all the applications that are sitting at a table in servicenow i can simply direct that base word to that table and not have to manually add all of those different vocabulary terms so that was a big enhancement our customers were asking for and then what we're going to show in the demo today more of a seamless virtual agent nlu authoring experience we have moved the the model um training and tuning capabilities as well as the edit capabilities once it's been built into the virtual agent authoring experience so we will share a little bit more on that today as well as the multi-language support for french and german so those are two new languages in addition to english we will have available with nlu now a couple things to mention on the nlu inference service this is really you know the external prediction engine for nlu so these updates have been applied to all customers but as far as the the mid-july release and there is a post on this giving more information on the community we've added some enhancements so you should see increased prediction accuracy basically what we did is added some technology to help the predictions understand the sentence context better overall accuracy so this is a really positive enhancement for customers you should be able to see some some increased prediction accuracy in your model although you may need to retrain it and if you find that you do have vocabulary pointing to other vocabulary words so in other words a circular reference you may need to go ahead and fix that and retrain the model and then republish it so again more is available on the community um but it's it's a great positive update and you know we should should increase the model prediction and so again multi-language support for french and german with the prediction server and finally the store apps so we will continue to release apps available on the store um so what you can expect starting in q3 is what we call a custom chat integration framework this is really the bot to bot framework a lot of customers were asking about um what we're finding is you know in bot ecosystems customers may have a main bot that they want to use on their intranet and they want that main bot to call out to secondary bots that might be application specific so what this framework allows you to do is connect the servicenow chat bot to that main chat bot and provide a seamless experience to your users additionally we are going to have an sms service channel so this is the the use case supporting a text to live agent for support so it is an asynchronous text to live agent and basically the use case for this is um you know if you have users that don't necessarily have access to a portal or not using virtual agents you know regularly sitting at a desk um oftentimes you know they still need to reach hrit for various reasons and so what they can do is simply text in um get support from you know the the service delivery side and feel like they are supported without having to ever go to a portal or access you know any internal infrastructure to do that so um again you can expect that in the store and then just to call out some planned spokes via integration hub for hr service delivery we recognize that a lot of the the hr conversations may extend to other applications and so you know we plan on releasing spokes for the various applications so that you know as part of your dialogue and part of the virtual agent experience you can use those spokes to build interactive experiences connecting different systems with servicenow and then voice of the employee so this is really you know what we call a pulse survey and a way to measure and gauge sentiment for employees this is really exciting it's a way to understand you know how are people liking these new experiences that are being deployed how are they responding to them are they engaged so this is really designed to address those types of use cases and then in the future you can expect additional itsm virtual agent topics to be available via the store so before we jump into the demo um i want to ask are there any questions from the audience so marcel we had one question that we answered and was is their dynamic translation supported in virtual agent um so currently the dynamic translation is only available for live agent and it is on the roadmap for virtual agents great thank you nabeel um okay so let's jump into the demo because i know you guys are dying to see what we have for paris um so i'm starting over here under collaboration because collaboration is allowing me to show you all of the different branding menu chat setup as well as the virtual agent designer but let's take a look at the the authoring experience within the virtual agent and so what i've done is i've created this topic ahead of time and i've also gone into the studio and built an nlu model ahead of time so i titled that virtual agent academy um we've created a new authoring experience with nlu but you need to remember the initial setup of the model still needs to happen in studio and so these are the assumptions you're going to see when i i go into this topic and let's go ahead and open up virtual agent academy so you'll notice it looks a little bit different in here right because we've added some enhancements first and foremost the nlu authoring experience has been brought into the virtual agent designer view so you can see because i've already set up this particular intent i can see all of the different utterances and i can add utterances in here so if i'm running tests and you know i'm previewing the topic and i notice um in my list of test utterances it's not recognizing um specific utterances that i think it should in that case i can add them and i can train and then retest the model so that's one of the great enhancements we've added for paris we've moved properties so you can see we still um retain the same name description type etc so under type i can choose topic or topic block and i still have the ability to select a category and at the bottom we've moved the nlu model into the properties area so you can see i've selected the model and the associated list of intent we've also given you the ability to refresh this to take a look at those intents additionally the languages have been moved into the topic design as well so if i wanted to you know maybe have this topic in french i can simply select the french model and then select the um the english model as well so the corresponding english model so this is where i can set it up on the french side i can do everything from entity mapping and making sure the model is all set up here but these are just some enhancements we've moved on the nlu side so back to the flow you can see that i have this topic set up and if we test it out you can see what the user experience looks like so first notice that i have the ability to test it including topic discovery or to not include that and the testing has expanded to include variables and analyze test phrases so i'm just going to enter a simple query one is virtual agent academy and it's showing me even in the test preview the accuracy match so i would previously have had to done this and had to do this in studio um i can select yes i would love more information on academy um so great it's every two weeks and what topics would i like to see covered in future sessions so internationalization you know comes up a lot so i'm just gonna say translations and i've entered my query so we've concluded the chat so that's really the new authoring and the test experience for the paris release so it contextually gives you a lot more information um i will take a pause here nabil do we have any questions in the chat or does anybody from the audience have questions so far so there are no new questions but please if you have any please drop in the chat we'll be able to answer it even if it's about another feature feel free to do it so um know myself back to you okay great so we have an additional topic um that you know highlights the validation capabilities and i know this is particularly important for hr customers as well as it customers and so i wanted to show you um what we've added here in terms of validation so you can see we have you know the ability to support text validation email ip address phone number url and then a custom one so you can even define your own custom validations here which is um pretty great so if i test this and what i'm going to do this time is remove the topic discovery [Music] and it's asking me for a phone number so let's just say okay i'm trying to update my phone number and i'm just quickly typing in numbers and you can see this most likely does not match a global phone number so the system does not um should not like that entry um but basically you know the the format validation would ask you to start with the country code so it would say you know this is not a valid format or if i were to type in an alphanumeric code um so really that's you know the the design of it to make sure that we're validating you know email addresses etc so that the data integrity is preserved so going back to our menu before we go into some other areas i just wanted you to to notice um some additional enhancements so we've added the ability to retrieve your custom controls here your topics your topic blocks and then a new section called created by me so if you're specifically working on a subset of topics you can certainly you know search we've expanded the search to you know bring in category nlu language supported channels you can see sms here so if you're using the sms adapter which does require a twilio spoke and adapter but if you're using this you may want to build your conversation specific to sms so you might want to say press one for yes press two for no um so you have the capability to say is it you know default sms etc and then just the normal sorting so those are just some additional enhancements and one thing to note under general settings where we've added the language support here is this is where you would need to make sure that enabled equals true for those languages to render this is a really important step and so if you're testing this out and you're finding the language support is not working just come in here and make sure that this is set to true moving through the dashboard will have additional tabs so you can see that this is covering conversational data but you can also see if we've expanded it to cover things like topic user nlu so this is just broken down to be a little bit more specific previously we had the indicators on one tab and this is designed to be a little bit more specific so based on conversation topic even by users so how many unique users etc lastly some of the branding enhancements have been extended to include the multi-bot profile as well as some different options available in the virtual agent chat client so this should look pretty familiar if you are used on the orlando release or you've been you know working on branding on the orlando release but down below you'll see some additional menu items that have been added and this is where on that vertical ellipsis menu on the virtual agent chat you can add additional icons you can you know have different menu items available um so you know there's been examples of links to other sites so let's just say you know you have a an external portal or you have like a procurement site that people can order work from home technology from you might want to have a link out to that to say okay for all orders please you know navigate to this site and hopefully deflect some cases around inquiries that come in for orders so this is you know a really great way to you know give users enough information so maybe they don't even need to initiate a virtual agent conversation they can call someone or even go out to an external site from here so really we've covered a lot of different areas we've spent some time in the virtual agent designer we've looked at some new utilities in the virtual agent designer um specifically you know when we covered the flow area you can see that you know we've added the custom controls here so that's exciting as well as the file picker as opposed to the image picker and the descriptive carousel so we've covered this we've covered the dashboard some branding capabilities the nlu seamless authoring experience as well as um the language support for french and german so i'll take a pause there what questions does the audience have or does anybody have any comments or thoughts about what's new in paris so marcel i'm looking through the chat um will the new features of this release be covered in the va training as well where people can get the va micro certification so definitely yes um anything we show you anything relate to past gonna be covered in the training um the [Music] i see i see another question would you mind sharing the doc the service now link to enable dynamic translation for a live agent yeah so we'll drop you that in the chat um we have a comment from mark the new dynamic greeting topic is also cool it is definitely cool um mostly excited about it and we have another question will chinese be included in the future releases yes marcel do you want to talk about our roadmap for languages yeah so we plan to support additional languages starting with the really with the quebec release um we are looking to really expand the language portfolio so by the rome release we're hoping to have you know upwards of of 10 to 15 languages supported in different capacities um but you know we would love to hear your language requirements so i would say you know please contact your account representative and just um have them connect us with with languages you need supported because we we love to hear from our customers where we can help support them i think that that's all the question we have for now okay great thanks for everyone's time today thanks for attending and we look forward to seeing you in a few weeks
https://www.youtube.com/watch?v=ph_tFjieP9M