I'm wonderful we want to welcome everyone again today and we'll get started shortly we're just going to allow some time for our other participants to join us and this is going to be recorded it's going to be in YouTube and I will be embedding the YouTube in that community link shortly and those are do you need the full 30 minutes or more than that today no why are you actually expected to be a little bit less than 30 minutes just want to make sure that we stay on time scheduled since we're just one minute over the hour so I want to make sure again this is being recorded I hope everyone is enjoying their knowledge sessions I think it's fantastic that we are recording them for all of our users to be able to engage in it and get the content that they want when it's available to them or so they can do their own time schedules and I think that's been one of the positive benefits of going digital and again just make sure that everyone knows that we have we are recording this and it is going to be available on the community link that I've provided in the chat well we're two minutes do you want to go ahead or do you want to wait um let's wait one more minutes and then then we'll get started great I wish I had some groovy music queued up but I tried that in the past it just kind of complicated the audio well the first time is work right yeah yeah I guess I could put someone I think I have we go now we have more more guests coming so that's great all right let's see if I can do this all right well we are three minutes up to the hour so let's go ahead and get started and welcome everybody to another knowledge event let's go take it away thank you Lisa and welcome everyone at the success meetup session on how predictive intelligence is helping us in creating the tickets incidence cases final agent workspace or the service portal my name is Valtor Duvall I am a surface now certified master architect and working for fruition partners in the Netherlands I'm implementing these ServiceNow platform for more than ideas and I've touched different products in the ServiceNow platform like IT same HR security operations but always had a strong focus in surface portal and new innovations my current project is implementing a service portal with a large service catalog and knowledge management with that service portal also looking at the user experience looking at the virtual agent but also looking at predictive intelligence and how that is improving the user experience a not only for the service portal but also for the agent workspace for today I've set up a small gender with five topics talking about the goal what do we want to achieve within the project with using predictive intelligence and how can you start with that and then I'll talk a little bit about the concepts of predictive intelligence and about the baseline prediction flow so you at least have a good idea of how to predict predictive intelligence works within service now and then I'll show you our example of predicting a business surface or a category based on the short description and description of a incident because there are many sessions about predictive intelligence right and how to set it up but I really want to have this session focusing on get get it done and really showing that it's working for us and also enabling you to do the same thing so finally talking about the next steps what what else can we use within prediction predictive intelligence besides predicting the category or surface all right so our goal and what we wanted to achieve is that we want to use predictive intelligence to automatically populate the fields during ticket creation we want to do it based on the history of the incidents that we have and I'll talk about that history of incidents or any record time and if you work with HR cases or see you same cases you can do exactly the same thing as long as you have a good history or amount of data you need of course and we're doing this because we want to improve the user experience one of the challenges that we have and many customers are having that if you have a service model and a user and end user needs to provide the surface that is related to his issue or maybe during the incident creation or case creation you ask the user what is the category of your incident right well why are you asking good question because the user doesn't care about that right he just won't here's a shoe created or issue solved let's say so we're using predict intelligence to improve the user experience for T and user but also helping the agents in the agent workspace reducing the reassignments because using that history and using predictive intelligence we know based on on the history we know what the assignment groups surfaced or categories should be so improving it and that also brings us to a faster me time to resolve an ebony issue said the right team the right information available and may be providing solutions and already based on previous created incidents but ago I want to show today's really focusing on populating the business surface field within an incident creation right because that is a challenge that we had and I want to show you the depth can work very easily and it can be a good starting point to start with predictive intelligence the overall concept of predictive intelligence say well you need about a hundred thousands of requests I draw a hundred thousands of requests in your instance so you know you are an existing customer and you have lots of history data in your instance well lots of tickets are generated each month so you want to use them by using that set of data and by the way minimal required of tickets is 10,000 taken incidence HR cases right but a good starting point is if you have thirty thousand incidents you can get a good prediction the maximum by the way is three hundred thousand incidents one so set the right filter based on you might wanna if you close the incidents for the last six months and that is a good number you want to use that data set the data set is used by particular intelligence and really that moon with strange and knowing based on which keywords based on which sentences to which surface it should go or if you use the standard service now flow says well you can predict the category predict the assignment group or predict a priority or incident right but I'm going one step further and say I don't want these three well maybe in the future but today I want to show you when predicting the business surface and we do that with the same concept and the same application called predictive intelligence in ServiceNow so how does that work in the baseline if you activate predictive intelligence and to want to predict that surface when an incident is created other words is the user or the agent provides a description of that incident because we have that based on the description on submit a business rule is triggering the prediction API so that means you will have to submit your incident right really save that incident and after saving the prediction will be triggered and will take out the sentence of the description and calculate step two the prediction model then comes back with a confidence level saying based on the sentence and based on the history of your tickets it should go to the email category anymore this is all happening after you the update right so in case you created for this service portal user doesn't see the surface yet until he's safely whereas in most search now implementations especially existing serves now implementations they always had a category field or a surface field on the form and we expect the user to fill that in so that's happening on the front end so what we are doing within our implementation we were also bringing that to the front end to the form itself so what I'm going to show you later on is that we're not doing that after the safe or the submit but we'll bring that forward using a client script and triggering that API client side really in the browser itself in that case we can also verify the prediction with the user itself say we are dis confident that it should be this category or this surface do you agree or do you still want to change that so that is something we're really seeing the benefit of and that's something we have created in our implementation so let's have a look at the demo of doing it achieving the prediction of the surface based on the Internet's and therefore we have set up a solution definition within predictive intelligence for the sake of time I just took the screenshots and a video instead of going live into my instance so when we have configured in the instance in the solution definition so we've created a classification template whereas we have the output field set to the surface so that is predicting the surface in the incident table we're saying the input fields is the short description in a description but it can also be other fields right we can also say the solution there's a lot of text in there that we want to help to predict a better outcome but to keep it simple and to start right we're using a short description and description to then predict the surface as an output field and we are taking the data of the incident table where the active is false which means we are only taking the closed incidents and they are created within the last six months and so we want to have new data and the data should be accurate to use as well we don't want too old data to use in there and you can see based on the filter I have 45 thousand tickets which should be okay to do a proper prediction at least you start with it to play with and to help you get going because after you train this if you click on update and retrain into train which will take some time of course because really going to that prediction server within the same data center and creating that prediction model before you actually implement that I worked at and bring it to production you have to see if their training model is good enough or you would have to make some tweaks on the thresholds that how confident should it be but that is a simple solution definition so let's have a look at a short demo from the agent workspace and what we have configured is that we have a cool and an incident where schools will a type of task so we've changed the order of the form I'm really saying well we're starting with the color the business user and then providing it so description and description and then all the other fields are shown because based on the description we predict the surface and you can see actually when I pass it is that we have provided the short description in the description and we're still in that description and starting typing right now based on that short description using the agent workspace you already have the agent assist but then showing you related knowledge articles based on your short description that might help you solving the incident order taken right away but if we leave that description field my client script will be triggered like my own change launched it will be triggered and based on these two fields the descriptions of the incidents we are going to fill in the business surface but so that is something that you will see happening bless you saw in my earlier slides and look at the strips now baseline we can also predict other fields because I if this is your description based on the history of tickets the impact or the urgency should be this of course you can change it later on but there's a starting point this should be it think about the Simon's group in this case my Islamic group will be filled based on my surface so that is something depth that's actually the reason why we say we want to have that surface predicted and not the assignment all right so let me continue playing this you'll see when I put my mouse outside of the description field that client script is triggered and the business surface will be filled in so you can see it is automatically set to office 365 outlook team based on the description and we even provide a confidence level to them saying we were successfully able to select a business surface and we're 99% confidence that it should be dead business surface of course we are using he predicts an intelligent right so we are probably never a hundred percent in the prediction so the Jews you can always change then afterwards we are filling it in but you can change it just to make sure that all right we are making the prediction we are not on a percent sure right so you can override it and that is the agent workspace and helping the agents in selecting the right surface we have a similar thing for the surface portal because you can also create a incident or any other type of ticket fire the service portal and we really had a challenge for the end users to select the right surface so we said this is definitely something we want to have on the service portal so let's have a look at the demo of this service portal where we are creating a new ticket we start of course with the user open on behalf of but then the next two fields are the short description in the description and we are doing exactly the same thing so when I change the description and remove the focus these surface will be populated and on top of the screen there is a message with a confidence level the future we probably shall we get at a field level right so it's more official for the users at this moment it's just and ServiceNow in four message but if you can change it because we're still in an early phase of implementing predictive intelligence we're not on up sensor so again the end user before we had predict intelligence he had to select it right nothing will help him selecting the right service actually so we are now moving forward with predictive intelligence and are really improving there and after each training we are improving the prediction as well and make sure that we're really helping that user in selecting the right service and this can be achieved very easily right and for the technical person in the meeting or maybe you are what saying is back so so you can have a look at the script as well and this is not very well documented or reduced not to be very well documented right of course you have that business rule that is triggering that but if you want to use that from your client side how does that work so there are some API available that I just wanted to highlight I'm not going line by line what it does but in case you want to do exactly the same thing I'm just highlighting that there is a specific API that you can use whereas you have to provide your solution definition from the solution definition that I showed before the videos and also providing the text that you're using within that prediction on the short description and in description that will then return your confidence level which you can show in the agent workspace or in your service portal and it's that easy to use actually and also that client state is just opening or calling that let script include fire glad Alex so that is something that anyone can do but of course the machine learning predictor API that was something that was not really good documented so especially wants to hire that in this session and with that you can achieve exactly the same thing predicting something based on the history and doing that on the client side and either doing that find a service portal directly on that form or the de record producer or you know any other form that you are showing or within the agent workspace so what are our possible next steps because this is nice right we can predict something based on the history and we're really saving that effort from the NGO sis or the agents they don't have to think about it anymore but of course that requires ongoing training and ongoing maintenance so we will have a separate data team who will look into the predicting data and need training models that will be littered looking at these thresholds and changing the confidence levels of each of these trainings so that's ongoing but besides that of course we want to use that for other fields as well maybe printing a category a priority etc which can be exactly done in the same way but there are other ways or other things that we can use from produce intelligence the similarity framework for example identifying similar incidents and propose a major incident out of that if you have six incidents that are created on that day which have kind of similar description my predictive intelligence will know that and will say well you might want to create a major incident clustering another thing that we can use clustering similar incidents and maybe identifying the problems that you can create so really helping that and seeing well these are my instance that are created in the past and analyzing that data then in Orlando there is a new feature introduced which is the knowledge the multi insights which will identify your knowledge gaps they will know your incidents that are solved and there was no knowledge article available so we showed that gap and efficient wise that to these knowledge creators so they know where to improve later on am i doing that we definitely improve the user experience but also improve our overall service that we provide to our end juices and I want to close up with that with thanking you for your participation and for joining this session I know it was a little bit shorter session there my previous sessions but I hope it explains very well in what you can do with predictive intelligence and how easy or how quick you can set it up to predict specific fields and save time from your end users so if there are some questions we do have time left for some questions otherwise this will be posted on the community and you can always ask your questions over there if someone is interested in the scape includes and client script feel free to ask that is the only community and I will send them over to you as I would like to thank you all and enjoy the rest of your day thank you so much for your presentation today and I'm not seeing any questions come in but again I'm going to reiterate that you can please post your questions later on that community length and you can also rewatch this this video so thank you so much and this presentation also have a great day I