all right folks um we are at uh 3 minutes after the hour thank you very much for joining our webinar today uh good morning good afternoon good evening wherever you may find yourself and thank you for joining the live on service now series uh today our Focus topic is now assist for CSM my name is Andre I'm the outbound product manager in customer workflow I'm joined by my colleague Banu and Abby uh who will support with uh question and answers as well so let's get rolling now no product manager presentation would be complete without the yay old Safe Harbor statement uh as a funny thought experiment you might try to hop into chat GTP and get chat g2p summarize the Safe Harbor statement as a Shakespeare and see what it comes up with but this is just a reminder that today's presentation is based on our existing Vancouver release functionality as well as our September 23 store release for now assist for CSM any future looking statements that may be made during this presentation will be explicitly noted but any purchasing or deployment decisions should be based on existing product functionality uh today's session is part of a ongoing series live on service now our goal here is to connect you with service now experts and peers that can help you products faster and realize greater time to value with the now platform we hope you join us for another webinar or the three on the 360 exchange you can check out the schedule by scanning the QR code or by accessing the link that has been shared via the chat here are some brief housekeeping reminders so please feel free to submit Q&A during the presentation we have a number of product experts that are eager and happy to help with questions we will also have Q&A at the end this presentation will be recorded and shared on the community and after the event please let us know how we're doing please influence the future format of these webinars by completing a brief survey so with that said let's get it rolling um so the speakers today will be myself and Banu once again outbound product managers for analysis for CSM so earlier in the year or maybe in the late latter part of 2022 generative AI started hitting the airwaves and it was a fascinating technology that will fundamentally transform the way that we work uh both as employees as customers whatever Persona you look at it from and today we're going to have a look at how we can put AI to work on the service now platform but long before generative AI we've been focused on AI in general here at service now and specifically what we might want to consider is what why does AI matter from a customer service perspective and it matters because it directly aligns with our prime directive here in customer service management and that is to improve customer experiences while reducing cost and a lot of companies believe that it's an or choice but here at service now we are keenly aware that you can do both of these things together you can improve customer experiences and reduce costs and we're focused on three main areas in doing so namely Improv in the experience of customers and agents increasing agent productivity and helping growing the agility of support teams and organizations so for over half a decade we have made strategic and intentional investments in AI this is going all the way back to Kingston if anyone was around in Kingston we'd be happy to know about that in the chat so let us know perhaps the release that you started using the service now platform on that would be an interesting data point and when we started our journey into AI we started with task intelligence and task intelligence is largely based on what we call or generally call predictive intelligence and what do we mean when we say predictive intelligence well predictive intelligence is all about identifying and predicting Trends in data by making OB observations about that data with pi you can detect things like similarity clusters of similar cases anomaly in cases and other generalized patterns and specifically with task intelligence we are able to predict things like a customer's initial and ongoing sentiment we're able to influence routing by peering into an email attachment and categorizing a case for example we're able to automatically ex ract data using document intelligence and we can also detect cases in 20 plus languages to influence routing to make sure that the customer gets to the most appropriate agent the most uh a person that understands the language that they speak in short order and to wrap it all up we made this very easy with a centralized One-Stop shop admin console this allows a admins to deploy AI Solutions with clicks not code and they can simply light up those AI Solutions and they could move from simple recommendations to full-blown automation as their confidence in the AI capability grows so that's where the landscape looked like from the perspective of predictive intelligence and traditional task intelligence so we have a couple poll questions here the first poll question is how familiar are you with task intelligence okay and we'll give it give it about five more seconds all right let's go ahead and end the poll and let's go ahead and share the results so we can see that there's uh vast majority of you are actually not familiar or somewhat familiar with task intelligence and this is actually not that surprising because there it is a feature that is deeply embedded within CSM and it is feature that customers tend to kind of Overlook so I think one of the points of this poll is to illustrate to you that there is already a lot of Rich AI functionality that exists within the CSM platform that you can leverage as a customer today and this is nothing to do with generative AI quite yet but we will definitely bring the generative AI point of view into the conversation so let's go ahead and ask another poll question what is your timeline for deploying AI Solutions into production all right we'll give that poll another five four 3 2 one and let's go ahead and share those results so this is also uh interesting and this is very common we find that a lot of customers are doing that initial kicking of tires when it comes to AI they're very interested in the promise that it brings but they're also a little cautious because they don't want to inject uh very transformative transformative technology into their workflow in short order so you know there's a lot of customers that do that three Monon um half a year kind of evaluation and then typically what we do is we we see is we see a swell of interest that occurs a little longer term when they start noticing the impact of it on the industry or perhaps it within their peir group and this is why this live on service now session is important is because you can connect with your fellow uh peer peer group members to kind of go through this journey uh together so let's go ahead and stop sharing that poll and let's go ahead and move on so thank you very much for that feedback and that's some very informative and insightful information there so that was our story so far right the industry was kind of buzzing along focused on predictive AI but as you know in November of 2022 all of this started changing with the arrival of generative AI now whereas predictive AI knew how to identify patterns in data generative AI added this additional ability to create new patterns from data in three simple steps first of all it understood the context and the intent and perhaps even the sentiment of the person making a request so for example let's say that you had a customer that was asking about a specific product or encountering a common issue and perhaps there they were rather upset with generative AI you have the ability to identify that context and intent of the customer the second thing that it did what it it you were able to go through multiple information sources and combine that relevant understanding together so for example generative AI can look at knowledge it can look at work history perhaps comments and what it can do is it can use that data to help triangulate the answer that it's going to generate and then the third step is that it can generate the content in a way that effectively communicates the findings based on step number two so what it can do is it can combine that information it can take that contextual intent and awareness and then it can generate a response with awareness of tone using perhaps specific technical jargon jargon and even with the context of that information in mind so with these three steps generative AI is really quite a GameChanger to a lot of the experiences that you may have a as a customer but also that the experience that agents will have as well when they're interacting with the system so the impact of generative AI was immediately clear to the industry but the manner of integrating it into an Enterprise grade application was not right gener AI was largely a consumer grade type of application and uh how was it that we were going to integrate this in a responsible manner to Enterprise applications now the thing here at service now is that's kind of what we do for a living right right we make the world of work work for people we are the platform of Platforms in a very siloed um platform landscape so it was very natural for us to take large language models and bring them in to the service now platform and we started doing so by creating domain specific models we have something called the now large language model which is used extensively within now assist for CSM and it supports customer support types of use cases however we have made our architecture very open you do have the ability to integrate public large language models if you do if you would like and this can include public large language models like open AI Microsoft ashure open aai and Google Cloud uh offerings in general so we have a very open architecture but we also have our own domain specific large language models to handle customer specific workflows um I'm going to pause on this slide I just want to go and refresh my demo environment to make sure that it doesn't time out so let me just do a screen swipe and come back okay uh but generative AI goes well beyond just the technology uh with any AI there is the importance of um having an ethical Foundation by which you develop Solutions and by which you practice business relative to those Solutions so here at servers now we have a responsibility to earn your trust and in doing so we have four guiding principles the first principle is that we are human human- centered which means that we provide clear guidance on how AI is used and exposed within any of our Solutions empowering customers to make a choice of whether they would use that AI recommendation or not we're committed to diversity and um bias so we're committed to um having diverse data sets and continually testing and understanding the impact and controlling bias in those data sets from a transparency perspective we have open communication through our product documentation we focus on the clarifying any limitations that may exist within the product so we're very transparent in that respect and finally we have trust in the sense of we have both internal and external oversight uh in ensuring customers have confidence in the AI capabilities that we roll out into production all right so with that Foundation laid let's talk a little bit specifically about now assist for CSM so analysis for CSM is designed from the ground up to help accelerate agent productivity we're enabling agents to gather faster contexts when they encounter a customer chat or case for example we're enabling them to rapidly generate summaries on cases and interactions and we're also enabling them to improve operations by collecting consistent data when they're resolving and wrapping up cases so these are the high level value propositions for analysis for CSM on September 29th we released five major features related to generative AI namely AI search Chat and Cas summarization a very Nifty now assist side panel and we also made uh Administration very easy with the now assist admin so let's quickly have a look at each one of these the first one on on Deck is the now assist and AI search and now assist and AI search improves the self service experience by generating answers in Portal and VA searches so rather than customers having to sift through search results we are now looking at the top related knowledge articles and generating a response for them and this is going to help Drive deflection and also improve the overall self-help customer service experience now if a customer is unable to get help via that self-service mechanism they may decide to interact with a live agent and this is where chat summarization enters the picture with chat summarization you can generate a summary at three different points when the chat is initially offered from to from the virtual agent to the to to the agent you can generate the chat summary you can do it during the chat with a quick action in an inline Manner and when the chat terminates you can also generate a summary this will help improve overall productivity for the agents and decrease wrap-up time so they can move on to other work that may be pending with within their support organization now as you know you don't always have simple one anddone chats support organizations often deal with perhaps longer running cases multi-touch cases cases that come in Via portal or via email with case summarization agents can generate a summary they can edit that summary and then what they can quickly do is push that summary to the activity stream so what the agents can do is they can rapidly gain that contextual awareness of that complicated case and perhaps take escalated action or extend Goodwill to a customer as may be required to provide the best service experience now any of those capabilities that we have chat summarization and case summarization can also be exposed in a conversational manner in what we call the now IST panel with the now assist panel you can quickly click on a skill or you can simply type in summarize a case generate a resolution note and the system will generate the associated response so agents can trigger these skills on demand they can also ask follow-up questions for example they could say hey can you make that resolution note shorter can you make it more formal or informal or can you make it more empathetic as an as an example so you can have those kind of more nuanced responses by requesting generative AI to change the tone of the response this helps improve overall agent productivity and provides more empathetic replies to perhaps sensitive customer issues and then finally we wrapped it together with an admin console this admin console allows you to deploy AI Solutions with ease and confidence with clicks not code you can configure skills you can toggle display options on or off for the workspace for example and in addition to that you have the ability as an admin to gain insight into how specific features are being consumed so you can see how often a specific skill is being executed by agents how many unique user visits do you have when launching a given skill you can even see the average completion time of skills over a given time period so with this admins can gain insight into the uptake and real world business impact of generative AI in the customer service organization so maybe I'll take a breath there I'll take a yoga breath and see if there was any I know questions are being answered in the background but is there any specific pressing questions that we need to um draw light to and I'll do that while I'm refreshing my browser environment so let's see if there's any questions if there's not we will continue and we'll get to the demo awesome so let's uh hop over to the live demo so for this demo uh we have three personas the first Persona is the customer Persona and again we are trying to improve the customer experience Drive greater self-service uh deflection rates uh essentially just reduce customer effort from the agent perspective we are looking to streamline the way that agents engage with customers give agents greater contextual awareness and then from the admin perspective we're making it easy for them to light up generative AI features and to track the real world impact of those features within their support organization uh I won't go through these functional Flows In detail but they are here for reference this will be covered in the demo but the first functional flow is we're going to go into the portal and we're going to do a search and we're going to show how we generate a generative AI response then we'll also show how that generative AI response gets surfaced within the virtual agent before we transfer to a live agent in the second vignette we will take that chat and we will generate a chat summarization when the chat initially gets offered to the agent we'll generate a summarization during the conversation as well as when the chat terminates in our third vignette for a longer running cases we have case summarization we will show you how you can generate a case summarization based on the previous work history and work note that may have transpired within the case and then how you can modify those give it a thumbs up thumbs down and how you can post those work notes the generated summary work notes rather to the activity feed of the case record in addition to that if you are looking to close the case you can have the system generate resolution notes and propose a solution so once again we have the ability to interrogate the back and forth conversation that may occurred with the customer or even within internal support teams and generate a proposed solution and surface that and populate that onto the case record form we'll have a look at the now assis panel which allows agents to gener to make ad hoc and On Demand request from an now assist and execute that against the specific case that they may have in Focus or in context and then finally we'll have a quick look at the admin console so these are just some functional flows that you can look at um perhaps after this uh webinar to review the demo so with that said let's hop into the live demo and if life is good let's see if there is no timeout and I will verify that you guys are able to see my screen where so now what you should see is you should see a side by side browser on the leand side I have George Warren who is a customer that's authenticated into a self-service portal on the right hand side I have an agent John Jason who is logged into the CSM configurable workspace and who is AA able to receive work so over here the customer George goes in and perhaps he's having a problem connecting to the internet so he types uh internet issues and when George does that we are doing two things we are searching the knowledge base and the knowledge articles that George has access to and we are kind of cating and generating from those most relevant articles a generative AI resp response now again because of our commitment to ethical AI we stamp this in water Market with an answer generated by now assist we provide the summary and the idea here is rather than having to sift through search results George is now provided a generative AI answer immediately now George can also get that generative AI answer via the virtual agent so in the virtual agent he can come in here and he can type in Internet issue and once again we will take that search query we will look for Relevant knowledge articles and if life is good we're going to be able to provide a generative AI response in line within the virtual agent uh uh chat here so in a moment here you'll see the response come back oh this happened last time all right sometimes the virtual agent flow needs to be at the right spot let's try that again so the virtual agent flow was able to generate the generative AI response based on the internet issue question that we typed in and once again we watermarket with the generated by now assist and we're going to go ahead and we're going to say that we want to ask something else so in this situation perhaps there are some back and forward dialogue that needs to occur but what we decide is that we need to speak to a live agent so when we type in live agent you'll notice that George is immediately popped over to John and John can go ahead and accept the [Music] chat so when the chat is accepted and offered to John we immediately get a chat summary of the previous interaction that occurred and then the C customer may type in some add additional information so the customer may type in some information about you know they recently upgraded their router to the latest model Etc and they can go ahead and shoot that over to John and now what John has the ability to do is in an on demand manner you can click the forward slash and launch what we call a quick action and we can do a summarization of the chat now this could be particularly useful if the chat was perhaps long running um in addition to that let's say that a third party agent had entered the chat they may want to generate a summary of the chat in an OnDemand manner you'll notice that the chat summarization is private so this is specific to John George cannot see it within the chat transcript they can continue having a back and forth conversation at some point in time the chat will need to be ended so we'll go ahead and we're going to end the chat when we've solved George's problem and when I click on and chat the chat will get terminated and this will trigger the system to automatically generate a summary so you'll notice that the short description and the chat summary field are currently unpopulated but with generative AI we will now go and quickly populate the short description and chat summary so we can drive a quick wrap up of this of of of this chat so there's the summarization that was generated by generative AI so that was our first situation with chat summarization the next thing we want to have a look at I'm going to slide the screen over to give us a little bit more viewing pleasure and I'm going to move my zoom bar out of the way sorry about all the extra highlighting here um don't like that okay so let's say that um George is now not handling a chat and when you think about chats chats are things that kind of pop in like popcorn they're quick oneandone type of scenarios but often in support organizations you may have more long running cases you may have a case that comes in Via portal at the on the Monday gets updated on the Wednesday and it's still pending on the Friday or you may have a case that ping pongs back and forth between different support teams so in this situation we may want to leverage cas summarization and here's a case the uh firmware upgrade on the router and you'll notice that in the middle of the activity feed we have a summarized button and if I click on summarize this is when where generative AI is going to look at the short description long description at the work notes and it's going to go and generate a summary for us in line and this is the summary that it generated so it's it it nicely defined for us what the issue is what the actions taken were Etc we can give it a thumbs up and a thumbs down we could also copy this summary to the clipboard perhaps we want to paste it to another part of the user interface but perhaps more importantly we can share the summary as workn notes so if we click on summarize those work notes we can see the complete summary we can modify or add additional notes to this summary so this is going to get F fixed in patch 90210 and when we click on save to work notes it will pop the edited summary onto the activity stream so in a few simple clicks and with a few simple edits the agent can rapidly gain that contextual awareness and rapidly wrap up that long running case now if this case happens to be an open case that's assigned there is a UI action button that conditionally appears it's called proposed solution and if I click on the proposed solution button then I get a modal dialogue box and in this modal dialogue box we will get a generative AI response so under the resolution notes this is generated based on the generative AI or analysis for CSM interrogating the work notes and the case itself to to to generate this response and we can say patch upgrade failed to be fixed on 1020 and we can propose that solution and when we propose that solution you'll notice that the case status changes to resolved and the case is updated accordingly so that's the second little vignette which is the case summarization the third thing that we can do here is we can also execute any of these what we call skills in an on demand and ad hoc manner so next to the search dialogue you'll notice that we have this little Sparkle icon for now assist we can go ahead and click on that and this will launch now assist in this conversational panel I could pin the panel so it occupies the right hand side of the screen and you'll notice that within here we can launch specific ative AI skills so we may say something like summarize a record I could just as easily typed summarize the record but we just made that very easy for you with a clickable button but we definitely do um recognize free flowing text so this is the summary that was generated once again I have the ability to like or dislike it copy it but perhaps more importantly I may say something like make it shorter and what generative AI is smart enough to do is to look at the summary that was generated and to provide me with a more concise summary I could potentially even say make it more empath empathetic or make it sound like Shakespeare or whatever the situation uh may be so this is our ability to expose generative AI capabilities in an easy conversational panel and you can launch those as by what we call skills and agents can do so in a on-demand manner and you'll notice that it's smart enough to pick up the context of the current record that you're working on when it is performing the associated action so I am going to end impersonation of John Johnson and I'm going to go back over to my admin Persona and the last thing that I want to show you here is quickly what does the admin experience look like so if you have a Vancouver patch to environment installed and you were to go to filter Navigator and you were to type now assist you will see the now assist admin panel and if you were to click on home it would take you to the admin panel from the admin panel this is where you can launch different plugins for the different now assist capabilities that we have now assist for itss now assist for hrds now assist for CSM we already have now assist for CSM set up so I'm going to click on this tab and I'm going to go over to customer and these are the two skills that we currently have configured for now assist for CSM in the future you may see other skills you may see other skills that enhance the virtual agent experience that's a forward-looking safe harbor statement you may see other skills that may be Rel related to knowledge generation once again uh that is a Ford statement there but in this situation what we're doing is we're going to have a look at the chat assist so we click on view details and we click on chat summarization and we can see how this skill can be turned on with clicks not code over here we can Define the trigger conditions for the chat summarization so is it triggered when the virtual agent handoff occurs is it triggered by a live agent action is it triggered when uh does it do a short description generation Etc we can also set where the skill will be be displayed will it be displayed in the agent workspace or even in ui6 will it be displayed in the now panel in the now assist panel in that conversational panel and if so with rols have access to that skill so you can control the security associated with with it so that's a little bit about the point and click manner in which you can light up and turn on these now assist for CSM features the last thing I'll show you here is how we make it easy for admins to see the consumption of now's skills within their environment So within the admin console here if we slide down we can see how the skill is being used over time we can see the other average number of unique users that are triggering that skill so with let's say we have an agent population of a thousand are 700 of those agents um executing and leveraging the nsis skill and if my demo environment had some more Rich data we would also be able to see the number of actions over time and indirectly we would be able to determine the amount of real dollar savings right so if that now is skill is saving two or three minutes per case and we have 100 agents times 300 cases you can quickly see the value of this solution in production and finally over here you have the ability to filter to specific scills and filter to specific date ranges so you can track the impact over time this is a very interesting poll question because um it's all often influenced on what your primary intake channel is and it's often influenced by whether your support organization tends to have quick one done cases versus multi-touch cases so let's go ahead and give this poll five more [Music] seconds 4 3 2 one let's go ahead and share that we'll go ahead and end the poll and we can share the results so uh with little surprise you can understand that the AI search and the virtual a agent experience is very critical that is often the entry point for a lot of service organizations that is a wonderful opportunity to reduce customer effort to drive deflection uh in an upfront Manner and also have real benefit in terms of deflecting work away from support agents so that quick oneandone work can be quickly shuttled away um as a little surprised chat summarization also comes into play so if you're not dealing if you're dealing with less oneandone work then what tends to happen is the complexity of the work when it enters the support organization increases so then that's where you would see case summarization coming into play 28% so thank you very much for that poll um some very insightful uh results there let's go ahead and stop sharing that [Music] poll guess I can close it and we can move on just to wrap up here so just rolling out of the demo we have some slides here that you can use for reference these are the feature detail slides which go into the um background of the specific uh features there are some nuances uh that you should be aware of but these go into the specific details about when when does the chat get triggered when does the summarization get triggered things of that nature I won't go through all these slides I will uh leave them for your reference for example it tells you which fields are used out of the box to generate a summary which user interfaces um a summary may or may not appear on uh so it goes through all of the the the detail there uh we'll leave that there for your reference we also have some some additional resources please if you're interested do download the now assis for CSM Store app there is also a now learning course for the introduction of generative AI so generative AI is a platform capability it is part of nysis for CSM but there is also available in other areas of the platform and we have extensive product documentation that goes through the details of now assist for CSM so this slide here um this is about accelerating value and further goals with these resources so um there's a number of resources that you can tap into uh including training and certification expert services and service now impact so please do check those out and with that said all right folks unless there's any other pending questions I think we can give you uh nine minutes back thank you very much for joining us today for this analysis for CSM session on how we are putting generative AI to work on the now platform and check it out in the store app