hello everybody and welcome to AI Academy as you know by now probably if you've joined us before U AI Academy is a great session for you but we it's not the only thing we provide you so here we have an AI and intelligence Community Forum you can access it with a quick URL it's Snorks you can go there and find all the resources for our AI products including task intelligence uh our colleague NE Ma just posted the task intelligence for itsm quick start guide I highly recommend that it's on the list of the articles on the community but also feel free to go to the community site to ask your questions and we have a very Lively community and different experts that can help you and and answer those questions so that you get all the help that you need uh if you familiar with now learning we also have a class called get started with AI Solutions it's a good overview of all our different AI Solutions and how to get started in your company this video is recorded and then posted in our YouTube channel that's the channel of the now community and on this channel there's also a lot of other academies and and videos that are posted very good resource to to get to and finally if you want to join in a partnership to share some of your data so that we can uh Advance the field of AI feel free to reach out to that email address for the global AI Innovation program uh and see what uh you can provide to help Advance our AI products as always in those sessions we try to cover what's existing in our products and what's available but in case you ask questions or we want to share some of the road map with you always keep in mind that we have to share the safe haror notice with you and if in case we make a forward looking looking statement meaning that we talk about something that's not available in the product yet you always have to reach out to your account team before making a final purchasing decision so that uh you are aware of what you're buying if you are new to AI Academy here are some of the basics of the academy it's content that we provide for you so we bring you fresh IDs to give a better understanding of our product and practical guidance as I mentioned earlier the session is recorded and then we post that on the YouTube channel for now community and finally if you are here with us that's a great opportunity for you to ask your questions so feel free to use the Q&A panel in Zoom to ask your questions and we'll make sure to answer those as we go uh the session today is uh organized with an overview of our product and then we're going to spend most of the time working through an exercise which is a a bit of of a demo of the product and guide you towards uh getting started with the product and then we'll save some time for Q&A and today with us is apexa to talk to us about task intelligence for itsm very very exciting and now aexa I'll give it to you so hello everyone this is aexa Dil from India so I saw a couple of folks joining from India as well so hello everyone good evening and good morning to all the people joining from us and good afternoon for the European folks out there uh so I'm the senior principal product manager with it service management U unit and I take care of all the AIML portfolio for itm core products here I'm with you to demo or take you through the task intelligence for itm as a first leas we released in third on third of August so when you hear task intelligence I know most of you would think that what is this is it different from predictive intelligence is it just the name which is changing or it's really something changing at a line so the good news is the task intelligence is quite different from predictive intelligence that's the reason we are doing these sessions along with multiple customers and how the TK intelligence is different the complete way how we build task intelligent use cases out of the box for our customers the way we think of about these use cases is different task intelligence is a purpose built AI for business users when I say that each and every word makes sense what those meaning is that business users mean it brings business outcome driven versus technology driven it means that we we don't say that hey you classification for this we said okay if we want to do incident categorization for all AI capabilities which we can use in task intelligence the way we think about the use case is totally different also the AI ml model which it uses is uses deep learning ml models which gives better results based on our analysis which we did in house based on multiple uh clone customer data that and it supports multiple languages long text attachment and so on so forth it also comes with a task intelligence admin console user experience which helps business users so when I say business user it means that the admins business users does not require to have the expert AI knowledge or they do not require to have a data scientist knowledge they just require to have their own business knowledge and then task intelligence admin console will take care of walking them through after each step and then deploy the model I will I will definitely demo that part today so uh that would be more clear apart from that the biggest value add it will bring to the existing customers of the artificial intelligences task intelligence support multi field detection model so you will see in my demo that I would be creating single model to detect multiple fields on the incident so it's it's very low on maintenance it's high on performance and it's very very easy to configure with all these uh benefits I'll now go to the exercise next slide please yep so this is the demos script which are the exercise script which I would be taking you through so consider a company where employees were not at all Satisfied by the service desk performance now the service desk manager analyze this situation and go to their business users and admin to understand what should we do because this reduction and satisfaction is also due to the higher increase in the mttr so this admin and business analyst decide to use task intelligence to solve this issue the task ad intelligence admin console provides human friend interface to train ass deploy and monitor the model so as I said earlier this admin and business analyst who do not have very great knowledge of the a IML can also deploy the model which they require for their business in the production once they do that they need to go to recommended actions framework to do certain configuration to show these recommendations on the service operations workspace once the service operations workspace start doing the prediction eventually what happened is it increased the agent productivity hence reduces the mdtr and at the end overall employee satisfaction score increased now you will see that hey schedule configuring something in task intelligence admin console then going to recommended actions to take care of how it shows on UI it's too much of configuration you talk about Simplicity but looks like there are two framework which I need to go and then do something the good news is we take care of task intelligence and recommended action configuration out of the box by autotrain model on your production what does that mean the moment you download task intelligence for itm plug-in on your production instance it aut to train the model for assignment group configuration item and services on your data provided you have the amount of data in your system and then also it has an automatic configuration in the recommended actions Shi out of the box so that just downloading of one plug-in will start the prediction in background mode when I say background mode it doesn't mean that whenever an incident gets created the prediction will happen but it will not show on the UI because you definitely don't want to surprise your agents until unless ad admin know that okay something is happening or you are allowing that so with this Auto Train model you will be able to start predicting in the background mode by just downloading the plug-in and how all these things works now I'll take you through the demo uh as I take you to the demo I would like to answer Julian's question so Julian asks that it looks similar to the predictive intelligence classification what's the difference here great question Julian I thank you that you asked this question as I said predictive intelligence require you to create three different models to predict three different fields but in task intelligence a single model will be applicable to predict these three Fields that's one the second is it supports multiple languages as well as it's very easy to configure as you will go through you will see the type of power UI experience it brings the type of power it gives to the admin is very much you do not require to have that technical knowledge to Precision recall which is there in predictive intelligence to understand whether your model is working fine or not in task intelligence it's very plain simple English so any business user who does not have AIML cap capability or understanding can also deploy this model so these are the two very big benefits which task intelligence brings with this okay so if you are on Utah instance so Utah 5 patch five and above or vancou instance and you downloaded the task intelligence for idsm plugin you will see this task intelligence for itm in your file Navigator in the task intelligence for idsm you will see two subsections or Sub menu one is setup and one is monitoring so this is also one of the benefit which you get with the task intelligence that you can monitor how your model is working so if you remember in just couple of minutes before I said that the auto training model will start with deploy your model in the background mode so in the background mode it will show you how it is actually performing without showing the recommendations on you UI so you can actually come here and say okay my model it was five times agents has accepted the prediction it mean and 21 times it has not how the prediction was replaced how it is doing and all those things you can see here even in setup we have certain matrices to give you a quick snapshot about your performance of the model but uh that in monitoring you can do a analysis okay so in the task intelligence for itm setup module you will see out of the box shipped this model itm out of the box task intelligence solution this is an empty model which we sh out of the box so that the moment you download this product plugin in production this model runs and get trained on your own data so that it predicts the value based on your own data automatically I will walk you through what this is is but for example you do not want to say predict these three Fields you can always go to this template which which we shipped out of the box and create a new model add your own fields which you want to predict and train assistant deployer model let me walk you through what we Shi out box so once you come here you will see the model name you will see the output table it means that on what table I need to train the data so my tring data is out of the box which we ship is the data should be false means all the closed and resolved uh incidents and I want to train this data on my reent incident so I just want to train this data on last 12 months what all all Fields I want to predict so I want to predict assignment group configuration item and service now what all Fields I want to give as an input for predicting this my input table is incident and the input Fields is description and short description this is also one of the difference our existing customer can knowe that in existing predictive intelligence we just use short description whereas task intelligence given it's D learning and long text support models the description field also can be sent as an input and we have seen in our testing that combination of these two actually provides far better results and then uh similar to predictive intelligence as somebody mentioned the number of Records which is required for you to train the model is 10,000 and I guess the customers who are know a knowledgeable about AIML know that these records the number of CS the more the maral is always better for your model to get trained so once we decide this we can launch the training and see how it's working so it shows that till today when I deployed this model in your past data which is last 12 months it would have predicted 47,3 and5 fields for you which is a great number to tell how you would have helped your service desk agents in entering the correct data into the system now we will see how the model is actually performing as of now so you'll see okay for the service it's performing quite better because 75% of the time what model predicted is what agent has actually return 25% it was deferred which is okay for assignment group I see it's 50/50% so I might want to go with just recommendation instead of autofill but as I see the service is doing far better I can say that okay I don't want my agent to manually fill it you can fill the uh service directly into the field and configuration item is also 4159 so I'll still keep at it as recommendation in case if you're not confident for example in case of configuration item that hey I don't want to uh do any prediction on UI I just say that hey do only monitor only it means at the back end it will predict but it will not show anything on the UI so let me show you all three for the exercise purpose that for service we are doing aill for assignment group it's recommendation and for configuration item because it's less than 50% I just want to monitor how my model is performing over the time so that I can come here in future if it the performance improves now I can save and continue it will give a quick snapshot about what all I have done and it will say deploy great so I'm already deployed it as I said you do not need to go to recommended actions as this out of the box rule has already mapped with the task intelligence rues so let me show you where it's get mapped so this is the resource generator of the recommendations of theed actions and here I would say that I want to predict these three fields and here when the resource generator is somewhere where I have ma this things okay till the time it comes let me give you a live demo the so let me create an incident in the service operations workspace so you will ask me that is service operations workspace is a mandatory for us to use task intelligence I would say no we can still go ahead and use our existing platform user experience with the task intelligence but you would not be able to utilize the power of how you want to show the recommendation it means whether you want to show as a recommendation or autofill and all these things happens in service operations workspace so I highly recommend strongly recommend that if you're are willing to use task intelligence you also use service operations workspace 3.0 and above okay so here is where I have mapped so for example this Con this generator type is mapped to this solution so this this Solution on the task intelligence is pointing to this Rule and then this rule is responsible to populate these on the service operations workpace and how service operation workpace show is let me walk you through this so I'll say email not working I will also say the morning I the moment I Tab out that's the beauty of service operations workspace which does not come with any other user experience the moment you Tab out please note that you have not created the incident yet it means that a service desk agent who is just typing in the short description and description the moment he Tab out to enter some other value he will be shown these suggestions now I want to walk you through what all we have done as a preferences in the task intelligence admin console the we said service auto poate it means that whatever is predicted I want to autop populate it here again it's just that I want my agent to have a quick view and also save their time so that's the reason I'm out of populating Agents can always change it if they want configuration item as the performance was less than 50% I said hey don't predict just do the monitoring in the background mode so that I want to watch it over the time so you don't see anything here yet and the assignment group as it was 50/50% I said just recommend so just recommending below the field that according to the a IML the assignment group should be C it's up to the agent whether to honor this suggestion or override it so these are the three things which you have predicted through task intelligence and in these three modes and also please note that the incident is not yet saved it means it's not yet created completely but AIML run in background and gave the suggestion so that even the incident creation process not just the triage process the incident creation process also becomes Swift and it gets created with correct data so that the incident get categorized in a right format that's it from my site from the discussion or for the demo purpose I'm happy to answer any questions thank you very much we have somebody uh in the chat asking again if it's only available for service operations workspace so if you could uh repeat maybe what you said about where it's available sure so uh as I said service operations workspace we strongly recommend that you use service operations workspace if you would like to use task intelligence for itm because you would be able to use the complete power of both the products in service operations workpace you will be able to call task in prence prediction even before creating an incident this beautiful power is not there in our native Native UI or the platform UI which you call one that difference the second difference is these these features where you want to show some as a autop populate some as recommended these two is also not used uh not available in our native UI but predictions are available so whenever an incident gets saved it will get saved based on the predicted value from task intelligence but that's exactly similar to what predictive intelligence also does with the native view of current UI so the beauty of task intelligence will definitely enhance with you use service operations Works Space great thank you there were also a few questions around whether we could use that in sub production uh so if you want to answer that and as well I would like to to provide one best practice around that but just to answer the question would that be available in sub production yes but sub production auto training capability would not be available auto training capability it means the moment you download the plug-in the model gets Auto Train on your own data this part is only available in the production but you can you can have this plugin on the sub BR train it manually see how it works and all that all right exactly right great the that's the one best practice I want to provide everybody here that we covered for it's the same for task intelligence for CSM it applies here to itm as well when you train your model in a sub prod environment really make sure that the data you have in your sub prod environment is refreshed is up to date is matching what you have in production otherwise you'll be training a model on on data that is not uh realistic and you would not get good results in when you move the the modal production yeah AA we have two more questions in the chat uh would this work with universal request great question so as of now uh this would work only with the incident but I would take that feedback back and see how quickly we can bring task Intelligence on universal request in our road map great thanks and another question came in which algorithms are used for prediction by TI sure so as of now task intelligence uses enhanced classification in the back end so that's the reason we as you see we shipped our first use case which is the categorization of the incident uh form as of now but going forward we would like to bring other use cases such as showing similar KBS similar incident ends also on Tas intelligence but as of now for our first release with in August we support this one use case and there were a few questions again when which release was that supported so you mentioned Utah could you give the patch ver version again yeah sure so task intelligence for idsm can be downloaded for Utah patch 5 and above and Vancouver and service operations were space 3.0 version would be required for task intelligence and the recommended actions which I demoed it the plug-in name or the store app called Advanced recommended actions for itm so by the way uh task intelligence do take care of downloading that but as a as a check I would recommend our customers to see that if uh they they have also downloaded Advanced recommended actions for itm as well then they would be testing this uh complete end to end product in their subpro orro anyway as ly mentioned nimma has created an awesome start guide in our community portal to give you an idea about the versions all the roles required and all I would strongly suggest to go through that absolutely I just shared the link for for those of us who are with us nilima is helping us with the questions today and she is an expert in that field so I really really recommend reading that article on the community well I hope everybody enjoyed the content today thank you apexa and Nima for joining us today and we'll see you next time thank you thank you everyone