so welcome to the session everyone as i said my name is janet acorn i'm your host i'm an itsm outbound product manager and i have paul van nissel roach adding with me also to assist and we're going to start talking about piloting servicenow virtual agent so just some host keeping up front we're going to keep you on mute and there is going to be q a after the presentation and demo please go ahead as we're going through if you have any questions at all add them into the q a feature in zoom to ask your questions and we'll keep an eye on those as we go through also feel free to introduce yourself let us know who you are so that we can reach out afterwards if need be and there will be some resources made available to you at the end of the session this session will be recorded and shared on the servicenow community so that you can go back and reference it or if any of your colleagues were unable to join they can listen in as well and last but not least there will be a short survey that you'll be invited to fill out right after the session ends we truly appreciate your feedback after this session in approximately two weeks time we're going to have a follow-up session which is about your learnings from today your questions anything that you've come up against in your pilot and so on to do a deep dive on jump-starting virtual agent and understanding how to move forward from a pilot and so on it's a really good opportunity to learn from servicenow experts to talk to people who have the technical knowledge and the experience and really also to engage with your peers so we hope to see you there we hope you'll join us back on april 26th for the americas that will be at 10 a.m pacific time or 1 pm eastern and for mia it'll be 3 pm gmt there is a link in the chat which you can use to register for that follow-up session so thank you again for joining us today my name is janet acorn i am a senior principal outbound product manager for it service management with servicenow and i have about 25 years experience directly in i.t in all different industries doing service management consulting both for other companies as well as to having joined servicenow about five years ago and paul if you'd like to introduce yourself yes thank you janet good morning everyone my name is paul von nisteloy i'm a senior principal product manager as well for our itsm business unit uh servicenow coming up nine years so a bit of experience here within this company and i'll be doing today's demo uh back to you janet thank you all right so for the agenda for today we're going to talk about you know what is virtual agent what types of conversations do you have with it and really what are the foundations that you need to put in place to have a successful pilot for virtual agent how do you scale that afterwards and grow your conversations we're also going to give you what resources are available to you to help you what your next steps would be and then as paul indicated he's going to walk you through a demo of a number of things including you know where natural and language understanding is turned on and how that's utilized selecting pilot conversations understanding what good topics are how to how to identify those and also piloting virtual agent with collaboration tools such as microsoft teams so that you're really working that virtual agent with people in the collaboration tools that they're used to using every day okay and we also want to show you how to create a great first impression with being able to customize greetings and conversations so that they're in a format that your users truly understand and are comfortable with so let's start with some basic definitions virtual agent is really it's an in platform conversational interface it uses automation to provide users with personalized and contextual interaction so today's virtual agent with natural language understanding interprets beyond keyword constructs to understand process and respond in a language that your users truly understand it means that you're able to utilize business language common terminology you no longer are left with only specific terms that are keywords and that are left behind from our traditional way of work you can now ask how to open an incident in about 50 different ways if i recalled number of times and it makes it very very user-friendly that way so that if somebody asks the question and types it in wrong forgets a word or something it still recognizes and moves forward so it's very user friendly virtual agent conversations support a lot of business goals it's all about scaling business efficiency reducing costs reducing the agent workload letting them focus on the things that actually take time and effort to understand and fix and allowing virtual agent to provide the information for very repetitive tasks and common issues it's going to improve improve service by providing users instant answers 24 7 so you don't have to wait for your service desk to open up to get your password reset and things like that it provides a lot faster resolute resolution time by giving self-service capabilities for piloting we really want to make sure that you identify your desired outcomes and that you ensure you introduce the big wins so one of the big things with supporting your business goals is identifying what is your top goal is it just increasing self-service or is it really driving that down that mean time to resolve in which case what you choose as a pilot topic is going to change is it a complex item or is it a whole lot of little simple items virtual agents support a number of different conversations there's requests for information things that focus on the process themselves such as what do i need to do how do i add an emergency contact how do i change my last name or request for knowledge that focus on the knowledge that you already have at hand in in your existing knowledge management systems and so on such as what is the status of my cell phone request where do i find information on new devices available things like that it also works for requests for action i need adobe installed on my laptop i need a new cell phone i dropped one things like that will be able to be responded to and start the request process prompting you with information as you go on so let's get on and start with how do you get there how do you start to get virtual agent working so there's three approaches generally one is starting with the natural language understanding and the out of the box conversations that are already pre-built at current time there's customizable templates for many conversations both for it for hr and for customer service scenarios for it itself there's approximately 65 out of the box conversations available to select from and that is what we will recommend for anyone starting a pilot that you start with approach one and use the out of the box conversations from there you go on to approach two which is really about creating and modifying those conversations with virtual agent designer so you have the opportunity to go ahead and copy one of the topics and then go in and adjust it to your own language your own number of tasks whatever it is if there's additional information that you need in order to move from step one to step two you get to go in and add that step in and work through and then the third approach is to create very custom conversations with natural language understanding models and the models really are just the list of phrases the utterances that you use and that your team uses and that are very custom to your organization so utterances can include some of the things like your internal application names and your project names and things like that that are very specific to your organization and that's where you kind of work to and you mature yourself to go from your pilot stage now slowly start making modifications to work with your processes and any changes and then the third one is really moving on and going ahead and getting into the details of everything that your organization needs from you know soup to nuts anything that is repeatable and has a common process can be added into virtual agent but it's really important to understand your user experience as you're creating virtual agent conversations so you want to keep it simple you want to have it fewer steps wherever possible and the nlu model or the natural language is really just to make it comfortable and easy to understand and ask so when you're considering the conversations for your pilot there's a number of things to ask yourself one what is your user's journey to really find answers where do they go to look for things you want to make sure that you have your virtual agent icon in a place that they sit in so whether it's on your service portal whether it's on your employee page that's where you want to make sure that your actual virtual agent connection is shown you want to be where the users are to get things started the second thing is are your users going to need live transfer are there cases in which a user wants to start a conversation and if they can't find something be able to jump out to an agent and just continue with opening a ticket and having the agent assist them further if so you're going to need to make sure you've got chat included or have agent workspace already running in your organization and align to which queue it will transfer third think about your knowledge base is your knowledge base going to support your needs for self-service do you have the content that's required is that content accessible to all users or only some users you want to start any pilot with virtual agent using knowledge that is publicly available that you've made available to all of your users because those are usually the things that are most common what feedback and metrics do you need to measure success so this is where you're going to say what is the outcome i'm trying to achieve and what do i need to receive to know that i've achieved it so if you're looking for self-service how many more interactions are closed that's the record from virtual agent as opposed to new incidents open so you should see incidents drop interactions increase right what else do you need to evaluate also is organizational change management available to support your pilot do you have champions in the field who are going to help people adapt to doing self-service and to asking va questions and just having good marketing around it because that is half of the challenge once people find virtual agent if you are using the topics that are relevant it will become very popular and will take a significant load off of your help desks next selecting your conversations as we said select some potential candidates look at the things that are always high quantity in your organization it's usually either changing passwords um changing email addresses adapting distribution lists things like that select some of those as your early on quick wins right conduct the analysis of those other things that are in your system and decide which are the most important to make available through virtual agent where you've got the knowledge to support it and take that load off of your resolver groups and then you select your conversations when paul does the demo in a bit he'll give you some hints on how you can identify these things and get the right information to start for a pilot keep adoption at the forefront of your planning make sure you provide accurate results to keep users coming back you don't want to have someone ask virtual agent how to change your active directory login and have them taken to something that asks or gives them information on new monitors you want to make sure that your language goes to the right things and all that means is testing so during your pilot do the conversation run it through get a small user group who you can you know bounce ideas off on how it would be asked and use that offer an intuitive interface get a group of users right testing to make sure it makes sense and then embed it where your users work so if your organization is using microsoft teams absolutely 100 percent get that team's integration in there early and make virtual agent available to them where they're working every day to begin with they ask virtual agent a question it can send a task out to another group in teams to do the work or it can respond to you right in teams it's right there where you're used to working and deliver a personalized experience again paul's going to show you a few things on how to change greetings and things like that that will show your users you know what time zone they're in you know what language they speak things like that that just make it more comfortable for them and prepare to scale this is the big thing make sure that your virtual agent pilot is in alignment with your organizational goals so that you've got the backing behind you to move forward and grow make sure you've got your success measures in place that you can identify how many active users how many virtual agent sessions you know has mean time to resolve been increased and build teams as you evolve to more complex topics start to work with more people to find out what should be in virtual agent what are the steps to be taken verify those things that are catalog items that can very easily be run by the virtual agent and create governance for your journey make sure you know what the technical standards are for building those conversations and how demands for new conversations are managed and agreed upon so that said just as an example before we go to the demo with paul virtual agent for itsm is really about improving your time to value right that's one of the goals and objectives so there's a number of different things in the new version in the san diego version in particular of virtual agent the conversation interfaces console which is shown here in the diagram helps you understand the outcome-based tasks that are going to help you update your virtual agent configurations and change things it's so easy to navigate and update greetings change your fallback topics things like that it's much much faster to manage and administer now you can improve incident and request diffraction by including the chat utterances as new data sources so what are the terms that users put in to va in order to get to the right topic and the right process to resolve their item where you have utterances that are not working you can turn them off and add new ones in right you're going to continuously improve those chat interfaces and be able to identify gaps with your self-service you have an analytics dashboard which is going to give you click metrics and know how many people went to va what did they go to va for what were the areas and one of the newest things that's really great for teams who have multiple servicenow instances is that teams integration is now enabled from one multi-instance to go to a single microsoft teams so if you have servicenow for hr and one instance and servicenow itsm on another both of those can use virtual agent and interface with microsoft teams in your single teams instance so it'll really help you work with users where they're used to functioning in their microsoft teams collaboration tool so with that said i'm going to pass it over to paul now to go ahead and give you a demo and walk you through some setup thank you janet yep thanks um happy to take over i think before we do so we would like to run a short poll with the audience to see where um the people in the call are in their journey uh partly out of curiosity and also i'll try and um take the outcomes into account in my in my short demo that i'll be doing so if you could um please fill out that poll and we see the results coming in thanks for those who have already responded almost there just a few more i'm not sure if everybody is willing to respond but uh yeah i think we're good uh i think we can share the results um so here we go sixty percent a little over sixty percent currently exploring options and 40 in the process of implementing virtual agent none of you have virtual agent live got it so that's really good to know and um what i'll do is i'll try and keep that in the back of my mind when i go through my my demo here um all right so as janet already uh mentioned this is the conversational interfaces home page uh this is a new uh homepage uh shipped with a san diego release i appreciate not all of you maybe on our san diego release the majority of the functionality uh exists in rome as well albeit that it has to be accessed through a menu and this is a way easier and condensed way um of accessing a lot of that information uh so um what i'll do is i'll run through some of the uh some of the features in this uh on this on this page and while doing so i'll bear in mind that some of you are in the process of implementing and some of you are still exploring options in my demo instance uh top right hand side i can see i have 30 active topics so i'm well on my way of implementing in this scenario 30 active topic is quite a lot actually especially for a first first goal it was recommend to start small and grow bigger but that's where we are in my my little demo instance here so i can see my bot on accident in a portal so if i have multiple portals maybe i have a knowledge portal and an employee center portal i can open up that portal and it will um it will show my my portal uh let's get rid of that poll and here i can see my virtual agent it's already popping up uh so that's where that shows in my um esc portal right so if it hasn't been set up there yet i can view the settings and i can add virtual agent to any of those portals i can try my bot from this page as well rather than having to go into the designer and go to test i can just hit try your bot and i can see you may have seen this um this window before it tells me now hi i'm your virtual agent let me know how i can help you today that's not very personal is it it's not saying me hi paul good evening uh and here's here's some of the information that i have for you it's just very generic let's get back to that later so i can either show everything these are the topics that i have access to or i can type in i am currently facing what i think is a fishing attempt and i would like your assistance with that so phishing attempt did not match any of the intents so what it means it didn't find anything uh any topic to match this too that's fair i mean some of the topics um we have um we have some um some some of the intents we have topics live for some some others not uh but it's not really giving me anything here so all in all not too good of an experience but let's get back to that well you know what let's dive right into that so if if i go to uh chart of to chat settings and i go to virtual agent right here's a lot of general settings that i can set but we're focusing on virtual agent tonight i'm going to skip a lot of these gonna come back to that later but here is my custom greetings and setup right so that greeting was not very personal and i didn't like that fallback topic either so if i open up my custom greetings and setup i can choose which experience i would like to use so you may be aware we are able we're capable of setting up multiple experiences for different target groups or for maybe a different portal and this is my default chat experience i only have one so that's easy choosing from me and here i can set up my setup topics and again you may have seen this on a similar screen this is not new functionality but it's very easily accessible through this home page and here i can see for my greeting i currently have a greetings topic a setup topic are topics that are reused in a lot of other conversations so that greeting topic now appears whenever a user opens up the dialog i want to change that to what i have set up as a dynamic reading topic so i can choose from a list right or i can start search here uh there it is dynamic reading topic i'm going to select that so this is a out of the box dynamic greeting topic that makes my greeting a bit more personal you know what i'm going to test that right away so hit save and let's find out if we can test that quickly so if i open up a test again there's my pop-up window again and it now tells me um what does it tell me hello system well system is not a very personal name but it happens to be that i'm the admin in the setup so we'll show you another example later on where it's actually using the first name but this happens to be my first name also 2500 open issues is not very good news but again let's call that uh a demo setup right so now it's personally giving me information that's relevant to me additionally i could say instead of hello i can say good evening because where i'm based out of the netherlands it's evening at this time or good morning if i'm based out of the united states right so i can make that experience more personal or i can for example say i know you're based out of the netherlands and by the way we currently have an outage on outlook so if the reason you open up virtual agent was to identify some issues with outlook you don't even have to type that in so that's really preventing or preempting better the end user to have to have to type anything in and very personalized to the end user so better experience the other experience that i didn't like very much was that fallback topic and a fullback topic is the topic that virtual agent resides to when it can't find anything i didn't find anything i couldn't match any intent to any topic etc currently to my fallback topic i could also set that to a search fallback topic again a search fullback topic is a reusable topic that i that is coming out of the box so nothing i have to do for that but i'm just going to change my default fallback to search fallback hit save and what i'm going to do a little the same test again see if it brings up any other result then i'm sorry i cannot help you uh if allswell is at least going to try to find something in the knowledge base so let's see what it comes up with first of all there's my hello system and you have a ton of issues and then i can type in fishing attempt exact same way i did before and what it will show me is one result of one so it couldn't find every corresponding topic but it did find that knowledge based article see so that's a better experience already and what i'm just using is out of the box building blocks to enhance my experience to improve my experience as an end user all right we're going to come back to this conversational interface homepage uh later on but for now i just want to just um take you through uh what an experience for an end user could look like um and what i'll do for that is i'll change my user to able tutor so i'm going to change users i'm going to pretend i'm able and able in our setup is one of our employees oh i'm gonna go to my employee service center i could have gone to my portal as well really the same thing here so what i'll do here um i'm gonna type in that i need software because i want to have zoom installed on my laptop and i don't need to i need a request for that but i'm not sure where to find it so maybe virtual agent can help me with that so what i'll do is i'll type in software because i think that's the best way of entering that and then it tells me okay so i found something software related uh request going to this topic flow first question is do you want to install update or remove software in this case i would like to install software then what it brings up is a list of potential applications that i could install now rather than this being hard coded or just a list that i typed in anywhere this is referring to the table with list of active applications right so this is dynamically built up if we decide at some point we're no longer supporting zoom then i don't have to change my flow or i just going a topic designer i had to change that it will automatically be picking that up because it's no longer an active uh an actively supported application all right this time i'm going to select zoom so now by answering a few simple questions it then tells me okay i will proceed to install software for zoom would you like me to proceed with this no because you know what i'm gonna try this a different way uh anything else yes i would like to just type in i want to install zoom on my laptop right so instead of just typing in software i think this virtual agent is a little bit more intelligence and maybe it understands sentences like this one now it's if you if you notice it's skipping the question do you like to install update or remove software it also skips percentage which would you like to select any of the following applications it automatically selected zoom this is what natural language understanding is capable of doing first of all finding that right topic so it knows that this is software related secondly having those entities like install event or delete derived from that utterance and making that choice inside that conversation so we don't have to ask the question anymore of things that the user already provided so i'm going to skip the question install a bit on the lead and also since it says zoom i'm going to assume that's the software that this person wants yep should i proceed with that now in this case nope no attachment needed in this case what it will do is it will submit a request so rather than me having to find the right catalog item and submitting a request for that catalog item through this dialog i'm now going to ask a virtual agent could you please enter that request on my behalf and you can see it did that and i can either use this link to open up the request in a new tab but for now this is good i'm happy with that and i can say nothing needed anymore now what i would like to do is i would like to switch gears a little and go to microsoft teams so here's a microsoft teams web application and what i can do as it says here the composite previous conversation is annette if you need help again type high so typing hi to my virtual agent let's say the now virtual agent here's either an epic once i've had my first conversation it shows you in my chest chat list uh this is just a demo instance which is why i don't have any other chats than just with virtual agent but i can wake my virtual agent up by typing in hi and i can i can see what i bring so first of all what it does it says i have three open issues i have one requested item and two incidents those were previously created this is probably the one that i i just just opened up so again i can use that link to exit out of microsoft teams and to go into uh services into service now but if i would for example do ask oops not escalate i.t ticket i'm sorry let's ignore that i'm done so wrong entry there for me not sure if it will let me maybe i should escalate this one let's go with it right so it has cleared that but that's not really what i wanted to do uh would you like to add a comment to it no i'm fine slightly off track here anything else yes i would like to get an update of my check i.t status see what else is there for me so two open incidents and one requested item which status would you like to check maybe that of the requested item okay i can either go to the requested item or i could add a comment here again i want to use a microsoft teams this is what i'm using this is what i'm working all day long so if i want to add some comment it would be nice if i didn't have to leave microsoft teams so if i would say yes leave some comment i can enter the comment any id on when this is a very level uh va level that's a comment that i added through virtual agent and virtual agent and added that to the requested item anything else yes you know what i would like to speak to a live agent so i can either type in live agent or i can select from this list live agent support oops click that properly hit submit your response was set please stand by while i connect you to a live agent currently the waiting time is approximately 29 seconds again i will have to shift gears opening up yet another browser in this browser i'm logged in as somebody called beth beth has set herself to be available for receiving incoming chats and this chat she still has 32 seconds to accept so as soon as beth accepts the chat that conversation through with a virtual agent has now been redirected from va to bath as you can see i can see the previous conversation so i don't have to start asking questions that the user already typed in in their conversation with virtual agent and i can start replying now so if i go back to how can i help oops or maybe because i know this is zoom related i can already find uh the request that was created and look into the request and maybe see if there's anything uh wrong with it or maybe i have an estimated time to when this will be responded to so if i say how can i help that's what shows up in my microsoft teams right so this is seamlessly transferring a conversation that started with a virtual agent to now a conversation with the live agent um right so i know really i'm done this is good so uh no need to follow this conversation through what i'll do here is i'll kill this conversation for now uh this will and the chat yep that's fine because there's one more thing i would like to show you guys and that is the use of um hang on my mobile app right if i see this is my mobile app the one that i'm holding in my hand here if i go to my work at the top i can access virtual agent and i can see the exact same here same thing here right so this is the latest conversation that i've just had i can see those three open issues so the same information is showing here show me everything i can check id status all of the topics are available it's just another way of entering the same information right and if i click any of those links i will get information i can add comments and all of the other topics that i have set as being active in my environment all right so let's go back to the conversational interfaces home page before i do so let me just quickly sorry and impersonation so whenever i need to go back into my service now that it won't be restricted to abel's authorizations right let's go back to my home page uh so you know at chat settings i think i was in the right section here um some of the things that um that i've skipped for now just went into custom greetings and set up the first bit here is topic recommendations you heard janet say that we have ways of distilling what would be the best set of topics for you as a as a customer to get started with and one way of doing that is by applying machine learning more specifically clustering framework to your data and by doing so we can look at for example the last three months worth of your incidents and then let's see if we can buy if we can cluster those incidents into groups that belong together and when i say group that belongs together what we do is we look at the short description uh we derive some some keywords some some some terms there based on the taxonomy that we provide and then we'll see if we can make some sense out of that and then based on those groups we can say these are the biggest chunks and let's see if we have an out of the box topic that would match that type of incidents now what that would look like is under the name of topic recommendations something that looks like this so again it's using machine learning technology to group incidents that i want to use now there's a lot of information on this page so let me take you through this the total number of incidents that i've run this clustering exercise for is seventy two and a half thousand this can be the last three months or a month of uh incidents uh we we need we need more than ten thousand we recommend more than three thousand incidents uh obviously this is uh more more data gives it gives a better a better view uh out of the seventy 000 going to my row here eight and a half percent or 6.2 000 are related to something that we think have to do with email issues right probably people saying i can't access my email or i can't i access outlook i have a problem with outlook or i don't know how to send an email all of those sentences that we think have to do with those those keywords they will be grouped into email issues right i get a bit more information if i click that i and i can see there's a mix in how people contacted us right 25 on self-service 20 over the phone uh 95 of these tickets were handled by the assignment group software and all of them were in english right so there's a bit of information here now what do i do with this there happens to be a out of the box topic that covers email issues so very likely this is a good starter to look into how to handle those incidents same with the troubleshoot slow computer or vpn connectivity right so that's 26 topics if you open all of them 26 topic that match somehow those incidents that we've analyzed in that discovery exercise now if you look at the first five over here so that's eight and a half percent at 14 that's 25 the 30 of the incidents uh could be covered with out of the box topics so that's 30 percent of the 72 000 and this is not not obviously it's demo data but it's not an unrealistic scenario right we can see that there's always a couple of uh large clusters that can be handled and if we happen to have out of the box topics for that that's a really quick start right so in order to get the answer to the question where should i start topic recommendation is really giving you a good idea of based on your data where you will probably find the biggest value in bringing topics live right and again five topics thirty percent that's a pretty good coverage right if you would extend that to ten you would probably add another five or six percent so maybe up to forty percent of the incidents that could potentially be covered with out-of-the-box topics that's a really good really good good starting point right so that's topic recommendations um another part that we that janet mentioned is the natural language understanding bit and i showed a bit of that in my exercise where i typed in i want to install zoom on my laptop i think i wrote something like that and it did not only recognize the intent what it is i want to do well something with software but also recognized installed and it recognized zoom as an application so this is through the through using natural language understanding if i go to natural language understanding or nlu workbench will give me a bit of a bit of understanding of how to build an nlu model right and this doesn't have to be huge i mean you need to have an understanding of how natural language understanding works uh but um once you have that understanding uh it's not a um a huge effort so in this in this model i only have four intents right i think i had like in my selection of lists there was like 10 topics so maybe i need to add to to add a few intents here to have a better coverage uh let's look at what intents i have in my model my model i have one for request software that's the name of my intent which is mapped to my topic software related request so there's a topic that i've built with va designer that's called software related request the intent request software in my niu model tries to discover when to call this topic right so when to direct the users to that topic and it does that by recognizing the language that the end user enters and by recognizing that it matches that to as we call that utterances now this intent only has 12 addresses which is a bit low therefore it says needs utterances so i could probably fine-tune my model by adding a few more utterances but let's have a look what those utterances are so here we can see my 12 utterances request software application installation request installation of software application and one software application installed now what are the these green phrases mean all these green blocks here those are references to entities and that software application if you remember that's referring to that vocabulary source that looks up into that application table do we still support zoom or in other words which applications do we support right so that's a reference to that entity installation is a reference to install update or delete so whenever there's a word installation or install or installed all of those words that's when the the topic knows to not ask the question would you like to install update or delete but direct this request to the install portion of the flow right so those are my utterances here i see my entities software requests self-replication that's my vocabulary source and there's a simple one with software so this is a this is very high level what an nlu model can look like the purpose is clear it allows for more natural and human-like conversation and avoiding having to ask things that the user already mentioned if i say i want to install zoom i wouldn't be happy if the question was would you like to install update or delete or ask for what application and by applying and are you to my va i can avoid those questions and make the conversation smoother what else is there here on this page one of the things is issue order resolution and with issue order resolution we can uh deflect incidents even after they have been created now that sounds a bit weird but what it means is if i am an end user and i forgot my password and i'm not aware of virtual agent or i don't think va can help me just creating an incident saying i forgot my password can be picked up by issue order resolution as something that we can that we can help the user with through the use of virtual agent so what it does is picks up that incident shortly after it's created sends out a notification uh to that end user saying i noticed you created an incident to reset your password if you want i can help you with doing that let me know if you want my support right and then if the end user says no i don't want your support then you can still stick with that incident or we can direct him to a live agent at that point in time but if he does then that's an incident deflected even after it was created so very powerful and it adds to the overall objective of serving the end user quicker as well as deflecting incidents and then this ai search ai search is a capability that we've introduced i think one or two releases ago so it's a machine learning based search which is a feature within the servicenow platform and it can be applied to a virtual agent as well remember where i was changing that fallback topic to a search fullback topic we can also apply that to an ai search fullback topic or we can apply ai in other areas so ai search can be integrated with or can be used within virtual agent as well right so those are some of the other aspects that once you get going with a servicenow virtual agent you can start to explore and pilot those functionalities as well lastly let's go back to my homepage um i'd like to just point out the new and improved branding capability so you may be aware that in previous versions we had the capability of changing the branding of the virtual agent popup or the console by using hexadecimal codes we can still do that but now we can instantly see what the result is so if i want to change that to bright red you can see or maybe blue you can see that that's the impact and this goes for all of the elements of that chat window now let me go back to the original color setting that back to what it was and one last thing that i would like to very briefly mention to you uh before i hand it back over to janet is the the analytics bit so analyze your boss performance through virtual agent analytics i don't have a lot of data in my instance here so i would like to for you to please understand that the numbers will probably be higher in an operational instance but just to give you some idea of the number and the types of metrics that we can we ship out of the box with this virtual agent analytics dashboard deflection count so how many incidents were deflected what was the deflection pattern uh was it through query response knowledge based resolve etc how much virtual agent activity is there number of va users number of conversations va success rate hopefully that would be higher in a real a real life environment percentage of topic flows completed 88 there's a couple of side notes on some of these metrics first of all if i see the number of incidents deflected uh i would also like to be able to compare that to an actually declining uh backlog of my incidents of those types for which we have va conversations brought life right so if i if password reset is now something we support with virtual agent i would like to see that number go up and the number of password reset incidents or requests go down at the same time so it's probably a few ways of measuring the success of those topics um let me browse quickly to the topics page here words perform the least used uh incomplete topics so these are incomplete uh that's probably me there breaking off that check i.t status earlier um most used complete etc so giving you a feel of are we on the right track is what we what we thought the end user was looking for the same as what they're actually looking for are they missing anything right uh giving you that insight in addition to that we always recommend especially in a piloting phase two to go out to your pilot users and ask for their feedback hopefully in a structured way but also ask them what what did you think of the experience what how how was that experience to you we have some really good examples of some reference customers that went all the way with that and results are always really good i mean once you listen to your pilot users and that's very valuable information obviously before before running this out with that i am going to hand it back to janet and janet over to you thank you very much and i will just reshare so let's leave you with a couple of takeaways for you to start your pilot for virtual agent number one is really to start with the out of the box conversations as we said there's a lot of conversations available to you you've now seen how you can identify which of those conversations are best matches for your past incident history and so on number two embed your virtual agent where people are using it so if you're using teams install the teams integration and make sure that you've got virtual agent available via teams don't make users look for it otherwise put it in the employee center so it's right up front number three align virtual agent metrics with your organizational goals make sure that you've got the metrics that matter that are going to prove the business value that you're going to get at a virtual agent okay when you're talking about which topics to pick just to give you an idea as we said there's over 65 out of the box conversations and there's a number of new ones so every month we add new topics based on what customers are coming forward with what's happening for instance if you're using citrix desktops there's a quick refresh so that people go out ask for that you click the button to say yes this is the virtual agent that i want to change what happens it will go out auto refresh and people can go on their way reusing that virtual desktop as if it was just deployed fresh right enrich the employee experience by really making sure that additional actions are there to manage meeting topics make it easy to schedule meetings for people have it go out and find available conference rooms whatever the case may be there's another one for making sure that notifications when software installation is available so add an actionable notification so that when sccm pushes out an application your user now knows that that application is present on their system and available to be used so basically you're going out and you're saying here you go we've delivered what you requested go ahead and go forth you can also complete the entire workflow for cloud provisioning with microsoft windows so if you need a microsoft cloud pc it can be deployed by virtual agent it's going to use the policies that you've set in place as to who can get a cloud pc when they can whether there's an approval and then no one else has to interact with it right so it's very very valuable and last but not least there's another new one which is using the process of requesting admin access so we use the itoms agent client collector to say okay who gets administrative access for what purpose what are the time frames what are the policies and a user can request it for a specific period of time so this is just a sample of a lot of the conversations but we're constantly adding on to these so if you go to our docs site you will always see current conversation lists available so what do you do next take time to review leading practice resources that are available on the success center we've got a couple of them here there's a virtual agent readiness checklist that is not just for a pilot but for readiness of her virtual agent deployment in general talks to you about all of the things to watch for the things to plan out some of it is much more detailed than a pilot but those materials will really give you a lot of insight from experts on what's happened what's worked and so on and then do your analysis make sure you have three to five really good out of the box conversations that are going to apply to your bottlenecks so that you have an immediate quick quick get those out there and start to work with your users so that you truly understand what their language is and their concept and how they work with things and you get that proper persona in place for your virtual agent right and then put your success measures in place to guide the decisions about where to expand and how you align with your leading practices including governance going forward your virtual agent can play a huge role in digital transformation in making it much easier and faster to deploy and resolve a number of issues so don't hesitate get it out there give it a try even if you only take 10 conversations in the first little while you're probably going to offload 20 percent of the work that's normally going to agents so again we're going to follow up today's session on april 26 with a meet up to bring all your questions your thoughts um challenges that you've experienced etc to talk to the agents and engage with your peers we hope we're going to see you there i know the link has been dropped in the chat for you and next if there are any other questions we've had a few questions along the way but if there's any questions please go ahead and drop those in the q a nothing i see there's been a couple that have been answered along the way and there's been a fair amount of chat so if there's nothing further i will say thank you and we will leave the q a open for a moment but i think we'll sign off and we hope to see you all on the 26th thank you very much for joining us today