before we get started a few additional resources we can find articles you can find the previous academies on our Ai and intelligence Community forum to get there one simple link it's SN dot Works slash AI and feel free to go to The Forum to ask questions and get answers from our experts as well you can find our recordings on the YouTube channel of the servicenow community and before we get started just take a minute to review our Safe Harbor notice and remember that if we do make forward-looking statements uh please check with your account team before making any purchasing decisions and uh just a quick review for those of you that are new to AI Academy some basics what is AI Academy it's fresh ideas that we bring to you to give you a better understanding of the product and practical guidance this session is recorded and then uploaded on our servicenow Community YouTube channel so you can access that later share that with your friends and colleagues and we do invite you to ask us questions today using the qnl the Q and A panel in the zoom interface and with that I'm gonna hand it over to yoast to get us started uh welcome everybody thanks for introducing us so my name is Jose I'm a solution consultant of the platform intelligence team and I'm joined with my lovely colleague Timo hi hi so uh I'm from Netherlands Steamers from Germany so we are neighbors and today we're going to tell you something about uh um well hyper automation so what do we uh think about Opera hyper automation within service now and also how you can use AI actually to kick start your hyper automation Journey how Ai and Hyper automation are uh well linked so um as we uh for our agenda today so we we actually this is part of a two uh series session so this is part one uh on I believe March 8th we have the second session and um today we'll have a bit more of an introduction on what we within service now I believe is hyper automation then how Ai and Opera automation are linked together and then we go into the exercise of um automation discovery the next session so part two uh we'll have a well we'll have less slides so to say and more exercises that we uh in March March 8th we'll go into clustering sorry clustering and process optimization uh at the end we have some time for Q a but if in the meantime if you have questions uh please use the Q a as Louis mentioned um in the zoom panel um first of all so hyper automation is we just want to give a quick overview setting the states um what we believe within servicenow is hyper Automation and it's actually linked to terms like digital transformation so we are in the next phase of digital transformation you you as companies you need to act fast you um um so you you there are new requirements for for uh to be able to compete or to move ahead as organization as a business and for that within within this domain we see three key Focus areas that that companies are paying attention to first of all hyper automation enables business businesses to operate at a greater level of efficiency and they do that by accelerating Automation and accelerating automation where it makes sense sometimes automation doesn't make sense we also want to stipulate that but you want to remove the lace and bottlenecks internally in your processes and those processes often are steered by internal systems Etc um secondly if you apply intelligence or apply artificial intelligence to those processes um you you can operate by well you can operate faster you can make smarter decisions and actions actually Guided by by AI and all of this and answers the day-to-day experience of our most valuable assets put in business organizations and those are people the so the employees of organizations so if you start empowering people with those Technologies uh and so they can focus on the strategy of your organization and the creativity that is within those employees so do work that they love instead of doing repetitive tasks monotone tasks that actually could be done for example by AI so that's those those are the focus areas that we that we see and actually if organizations start focusing on those well let's say three pillars um there are clear Financial results as well so if you focus on efficiency you would use operating costs if you apply intelligence you increase productivity so you can reach new levels that you couldn't reach before uh and you also will reduce risk so uh if you have very repetitive tasks so to say or monotone Dodge if you let them do uh or help AI with those tasks you will reduce the risk of for example errors and finally um employee experience if you focus on employee experience eventually well let's if you're if you're if the employees of your organization can focus on meaningful work uh they will drive Innovation for example and this will eventually bring growth of Revenue and also if organizations can provide people with with work that they love you will keep your employees so they don't go to your competitors that's uh that's a big issue as well they're still still war on Talent there you need to do more work with uh with less available people so to say a small example if we if we look to a simple procurement process like water to cash think about [Music] um in the in the let's say the old world um you you have a have a new order created but it's a it's a sort of unstructive process so it could be that somebody's uh emailing a PDF to an organization then somebody needs to see what's going on um third-party system needs to be uh checked for the the creditworthiness of this this customer um there's some some manual work there to check this and eventually you have to ask the customer for example by phone to to pay because the the whole process is done it's a sort of small uh Legacy approach and it takes a it takes a long time well if you would apply automation to this you can start thinking about presenting them presenting the customer with uh for example a low code app so give them an application where they can easily submit the order then when things like PDFs or attachments come in you can actually use document processing so like document intelligence this has come up in other AI Academy sessions as well you can extract values from attachments and use them further on in your process uh you can for example use RPA to to trigger influences and let's say connected systems that don't have um the the correct API for your for your system use use RPA for example um you can use AI for um well sort of checking the credit worthiness automatically in those third-party systems so let AI recommend if it's a good thing to start delivering on this order or not because uh the customer isn't or is creditworthy and finally you can think about digital digital payment is taken via portal and for example validated with with the app and eventually um you match the order to the payment and then the order is fulfilled so this is sort of um applying technology on an unstructed process to make it more streamlined streamlined faster and also applying intelligence and delivering a better experience in this case to the well also to the employee but also to the customer and what we say is um well actually with with our platform we we can provide complete hyper automation to any process with with a single platform so we we have different faces within hyper automation uh and we we always uh see four phases within hyper automation hyper automation is a sort of cycle of continuous improvements and well the question is where you start but let's say we have discover we can use the Discover tools on our platform to find automation opportunities or to find bottlenecks and processes so we can use process mining or we can use automation Discovery to find opportunities bottlenecks in the process if if we know this if we have those insights then we can start for example applying automation with in the build and digitize digitize phase so we can say okay here uh there's clearly a need for a low code app or here there's clearly some opportunity to apply let's say predictive intelligence or task intelligence to automate and apply AI and of course you do this um because you discover something you discover the bottleneck in your process you you saw with process mining this is typically part of my procurement process where there are manual tasks we need to improve there well if you if you start improving if you start building those low code apps predictive intelligence uh applying all those automations and intelligence to your process you want to know if you do a good job at it So eventually you want to see in the optimized phase okay I've started let's say in in March with um with this applying this automation intelligence let's measure in over time in June if I see an improvement we can again use things like performance analytics but we can also use process mining process optimization again to compare processes of time and then you see okay did we do a good job or not or are there again areas where we can improve and this continuous Improvement cycle keeps on going so that's what we believe is hyper Automation and we can we can support that with with our platform and where we are focusing today on is the the lower part of this wheel of 14 so to say so we're focusing on the Discover part but still if you if you look at this it is where where do you begin in your hyper automation Journey so I mean we have a we have a big platform there are a lot of tools available on our platform or also maybe outside of our platform you want to you want to integrate automates apply eyewear it makes sense um but therefore we also want to know what can we actually do with our platform so a quick quick check there um if we if we look at our platform and our different ways to represent this we have we have digital workflow so we've we have customer workflows employee workflows technology workflows itsm we can build a lot of things with creative workflows we have um the the foundation layer of the platform where we will there's a lot of stuff on security seem to be uh there are a lot of configuration possibilities to build low code apps um you can use Integrations to other systems but eventually where we today want to focus on is the intelligence part of uh of our platform so we're focusing on things like process optimization analytics and insights machine learning so we actually use that part of our platform to start let's say investigating where can we hyper automate and that's what we that's what we mean with we can use AI to kick start hyper automation we can actually use those Technologies of the of our platform the intelligence capabilities of our platform to get this Wheel of Fortune running so to say so if we if we look at the discover phase we we basically talk about uh these three capabilities so we talk about uh automation Discovery that's our main topic for today so you can quickly find automation opportunities uh most of the time if you do this in an itsm setting you can also find automation opportunities for the virtual agents but automation Discovery basically is part of predictive intelligence and it uses the machine learning framework clustering so we can also use clustering itself as a technique to find Trends in data and also calculates the the impacts of those Trends in data so for example you can find out in case data that you have a lot of cases coming in for let's say a particular service in your organization or there are a lot of tickets for VPN related issues if you know these insights then you can start addressing them right so you can start applying automation because there isn't yet anything to handle those incoming VPN requests and finally we can also use process optimization to well of course uncover bottlenecks in your process so we we can apply this on platform process mining to see where the bottlenecks in let's say the incident or any HR process or a combination of HR and IT process processes we can see okay in this particular stage of the process there it takes a long time or this is a potential root cause of things going slower um but finding those bottlenecks eventually also uh is means finding opportunities if you see a bottleneck it's you can convert it to an automation opportunity there is there's something that you can do better So eventually the target of this discover phase is to find automation opportunities find things that you can eventually solve within the let's say the build and digitize or the apply uh AI automate phase um and those those three techniques are are heavily focused on um artificial intelligence but before we go to in the artificial intelligence we always say that you if you start using those Discovery tools it starts with understanding so sometimes um you first need to take a look at what is going on in your instance before you start throwing machine learning edits and um we can do things like um answering questions like what products are in use how many tickets cases Etc are available in the instance which inbound channels are currently used um who uses the instance for example often uses login do we have multiple languages in use so if you if you start understanding or answering those questions and you can use the following methods here so you can for example uh analyze the answers and tables you can run reports with performance analytics is available you can check reports to answer those questions you can impersonate as a user for example and and check out how the portal is being used by certain personas anything that gives you Insight on what is going on and you can also just ask your colleagues or ask customer on on feedback of certain um certain processes and that's still not machine learning so if you start understanding and answering those questions you can actually use those answers for um those Discovery tools so if you have those answers you can start uh correctly apply automation Discovery or correctly start a clustering solution because you know where to look at you you know what to filter out or what to include in those clustering Solutions maybe the same comes for process mining you don't want to mine let's say the last quarter of incidents if you have two million incidents the last quarter you want to maybe filter that down to let's say 10 000 or 30 000 records because then you start seeing things and also mining millions of Records uh takes a long time and it doesn't really give you insights it will give you averages and averages don't give Insight so you want to specifically look at Parts there um so hopefully this this makes sense if you know what's going on you can use that as input for those AI tools but sometimes it also is a direct call to action so answering those questions sometimes can immediately say okay there's a thing that we can use because we see a lot of languages that are being used but we we see it we see a clear opportunity for example to install Dynamic translation let's let's let Dynamic translation kick in and handle some of these languages in use or it could be a direct use case for applying predictive intelligence so you don't need to discover that you can simply answer that by looking at reports or applying analytics at your data and just to give a give a example let's let's look at these numbers these are numbers that you can you can simply get out of your system without any AI so we can say we have 48 of incidents on a on a yearly basis we have 14 000 incidents average reassignment counts something that you can get out of your system so that let's say that's 4.4 and actually by the way this is a real customer example um we have 40 service requests and requested items and 12 generic requests um I can get this out of the system in like two minutes and no AI or machine learning has been applied with simply looking at at the reports and tables but this tells me something because if I start looking at this data and I apply a few let's say assumptions so I say okay 40 of the incidents um and that that counts for 14 000 incidents per year just imagine routing takes one minute so reassigning uh incidents from one assignment group to the other that takes one minute then approximately and we round up here but per incident that it counts to five minutes just for routing so if we start multiplying real simple math we come to okay five minutes routing 14 000 incidents per year you spent 48 days just for routing per year well you can you don't want that sentence that's uh time spent is money wasted in this case so this is a clear use case for example to apply maybe advanced work assignment or a predictive intelligence classification to automatically predict the uh the right assignment group we have a lot of data 14 000 per year if we are already there for a lot of years but then we have data that we can use to predict the right assignment group and you get this number of average reassignments down and that will eventually lead to a lower days for routing per year and you less time spent is uh money back in your pockets so that's that's the that's the use case uh for let's say not using always machine learning sometimes it's too obvious you can get it out of your system out of your system with reports Etc but of course um we also want to use machine learning we also use we want to use those tools and why do we use them um well a real simple example if we are looking at this data and just um assume that this this is the complete data of the system we see we have a system of six records and that's uh to think with me assumed that this is a full system then we can easily see Trends in this data so we know okay six records in my system I see that [Music] um five out of six have something to do with VPN or connectivity issues and only one that has to do with uh password resets well done the division is easily made we have 80 83 VPN related issues and 17 uh password issues in this instance um easy to uncover these Trends but that was because it were six records if you are looking at let's say uh 96 or almost 100 000 records uh this won't fly of course you need to you need to have some kind of intelligence to start finding Trends because you won't go over 100 000 or maybe 10 000 requests to see Trends and we can actually start using those techniques on the platform automation discovery clustering or process mining to start answering those questions and how to find those Trends in thousands of cases how to match Trends to automation of automation capabilities so if we see something happening uh can we automatically tie it to some kind of capability on the platform or how to find we are we have a lot of workflows on our system and let's say we have 50 000 records can we find bottlenecks and fifty thousand records you can do that manually but will take you a long time so that's where we apply machine learning to get those answers as said today we are focusing on on one thing uh and that is automation Discovery so one one uh one part one tool within the Discover phase and what automation Discovery is uh Timo will take you through later but it's it's about quickly discovering automation opportunities it's it's it's real easy to use it's real easy to set up it's uh it gives you direct insights it's as easy as running a report and um well you could I said you can quickly identify those automation opportunities and actually what it does it's it clusters the records that you send into automation Discovery into clusters buckets of um opportunities so you actually see okay I have a lot of VPN issues or a lot of Hardware related issues and what it then does it it will tie those opportunities to virtual agent conversations when you when you apply the itsm taxonomy if you and we also have a AI Ops or item taxonomy if you apply the item taxonomy it will um it will apply let me just go a bit further if you apply the um item taxonomy it will um it will give you insights in what you could have prevented with predictive AI Ops so you can also use this taxonomy on your incident data to see what you could have been preventing if you would have applied AI up so it'll it'll give you insights this these clusters you have available and by the way this is something you could have done to prevent those incidents from entering your system uh I hope this introduction made sense because now we are we are going into the exercise I don't know if there are already any questions coming in nothing in the chat now this is coming one can this be used for CSM um you you can use this for for CSM you can you can apply um the itsm taxonomy on CSM data if your CSM data processes is related to to it processes you can also choose to have no taxonomy and if you have no taxonomy you will simply cluster the data and you will get insights there as well so you you can apply that hope that answers the question which we have no more then I'll simply continue of course now we we are entering the exercise we have some prerequisites for automation discovery um there is a store plug-in there are several available languages so if you see your flag that's uh then that's the language that will uh that will it will handle there are some additional plugins where T-Mobile will go through we have a minimum amount of Records by default 10 000 but you can modify this if you don't have that many records in your system um and license wise this is uh you need a pro license like ideas and professional that being said I want to hand over to to Timo for the exercise perfect thank you then let us go to the non-slide part of this presentation and what I will do um first of all use can you see the instance yes you're good to go okay so whatever what I will show you first is how you would insulate normally so what you want to do is you will type in plugins you will open that and what I well it did this I open that here and if you're searching for automation discovery you can find it exactly here I already installed that on my instance I could now do an update which happens from time to time and also the other plugins that you just mentioned you can simply go and you can simply update if something is missing automation Discovery has a troubleshoot button which may help you um and because I saw that the question was the main difference between automation Discovery and top recommendation pursue we need to install also the itsm virtual agent conversation and a new model for that and with that this is the second piece that we can install so it's its um and then you monopolitual agent and that's the second one that we need to install again think about obtaining from time to time because the model could change could um increase the quality Etc so that's the first thing that you want to do is really to install that the second thing as yours mentioned is about understand so what you could do you could simply use now automation Discovery taking all your instrument that you've that you have in your instance millions or only twenty thousand or five thousand or what you also can do is you can first have a look how many you have and then more or less work on the filter mechanism that makes sense so for example in this instance and this is just our demo instance it could back sense to check how many incidents have been created the last I don't know what is in good value I'm normally just checking the last six months or three months um how many I have here of not analyzing tickets that could be two years old or three years old because there are potentially not describing what your business is currently facing what issues you have currently and looking for automation through so it's always good to yeah lower the number down and make it actual with for example working with the created timestamp so last six months last three months um you can decide on that second what I did here and based on my demo data I will just show all data it could be interesting to um think about a channel so if we have automation Discovery and we are looking for automation opportunities and I have automation Circle it could be that we are specifically looking for virtual agent and then it could make sense to start with the tickets that are already be um covered with self-server so it could make sense to check if the users already go to the portal then it could make sense to just checking those so this is easy you have a Channel or a contact type and you can for example just say Okay exit in another way show me everything which is coming from self-service and using that as a filter I'm working with that but for sure you can also start in another way and that's the reason why we already prepared some of those reports if you're not sure if you have installed something there's a troubleshoot button which will tell you um at least what is missing there are only some warnings that are not mandatory to install but you should install the taxonomies at least the vigil agent um conversation and a new model for itsm and if you want to create a new one that's also easier of your button here new report you're in going and you give the thing a name this is now my Timo test report for example you can choose a taxonomy and as you was mentioned we have our itsm taxonomy we have our predictive area of taxonomy and we have something called No taxonomy which is a a way to use clustering to start with a cluster activity which we are explaining more in the next session uh beginning of March I would choose for example itsm if I'm looking for your typical employee or end user related issues and this could also be used for whatever table you have um even if it's saying data source incident you can choose for example also customer cases here if you want um as and customer service case will be somewhere here so we can not only use that for incident you can also use it for different tables per task for request of finding automation opportunities what the taxonomy is an itsm taxonomy then you're going to select the filter as an example my created last six month um you can use and we're going to generate here the information how many um records you would use this will not pass because currently the system property is saying we need at least 10 000 um but we can also do things like I showed you Channel [Music] is um um self-service or channels portal or channel is I don't know mobile app or one off so you can work on your filters to modify the data source that we want to proceed with then we're going to ask what is the field that you want to analyze typically we're using the short description but you can also try to do it on with description so we need a free text field for that and then we are asking for open date and time resolve date and time open I guess it's clear that it's opened at resolved um here it could be closed or resolved and this is what we're going to use but the calculation of mean time to resolve so this is very very easy of having um really this the solution setup I will not cling on one report I will show you what I already prepared for you to go through that so if you're checking this um I was creating this yesterday those reports I'm just selecting that I needed once so they're not actualizing and making a new fresh version of it automatically um and I will open the itsm full which means I included all tickets in there all my 80 000 um with the itsm tax on me but it is now telling me how many opportunation opportunities automation Discovery found it's close to 150 automation opportunities have been matches that is what it's saying and 29 of that have pre-built vigilation topics so if you install the itsm virtual agent conversation and then a new model we can match those topics with this widget agent ready which means for example itsm issue Hardware troubleshoot we have a pre-built conversation for that that you can use the action taken as here for example just means it is already activated an action has been taken for that you can also use here the um information symbol here which is giving you more insights the same that we have in the second card is possible deflections and how many of those possible deflections that we're talking about have pre-built vigilation topics and it gives you again more informations about um opportunities we identified about almost 38 000 of the incident in this report of these polygraph percent can be addressed by privileged agent topics and the remaining seven thousands may be addressed with custom topics um but it's not only about topics if you find for example um VPN connectivity it could be a ventilation topic but it could be also a catalog item a knowledge base article it could be everything that happened on the platform to solve you know reduce those issues that is basically what we have let me see some opportunity volume so that we see also the records and the time to resolve that we have from the most common things action taken or not taken I'm sure we can also click in here and see some more values um if you're going next to the list um and then the list we have a couple of things that we can see we can see the record matches but only means in that example that 5500 527 incidents we can put in the Box called itsm issued Hardware troubleshoot um we can open that and receive even more information so based on my demo data I cannot calculate the mean term to resolve if you would run that on your instance you would see that you will see here your mean term to resolve your open to resolve or open to closed duration where you can see and you can check where do I should start where's the my biggest topic with the biggest mean time to resolve you see also assignment groups for service desks for it support Americas you see your top short description values and you see some your actions you can take some examples that we would propose what you can do with that and you can even go more into open topic recommendations and I believe we have a virtual agent Academy session which is exactly talking about open topic recommendations so not AI Academy we have more about topic recommendation in the virtualization Academy and we can go through more we can see for example we have a box called itsm issue software sip um where we're saying okay here are the sap tickets and or sap related tickets we have something for issue we have something for network but also we have something for item and with the item topic I wanna quickly go to my last bit here because this is our predictive AI Ops taxonomy um it looks a little bit different and just so that we have 16 opportunities one of the preventional potential is so predictive AI Ops could prevent that those things are happening or deflect those things with automation with 8.5 percent giving me the number and it always gives me the provincial mean time to resolve so this is not about a taxonomy that we are matching with NAU the content it's more about giving you ideas um how many you have here and with category database what you can do what is a reduction By Priority what is the combined mean term to resolve what does it reduce mean time to resolve so things that you can do and apologize for those values these are based not the demo data that we have available currently here last thing I'm going to show you is just think about also using this understand phase and this idea you can do that for all tickets but you can also dig in a bit deeper and slicing and dicing the data set and in this example it's just showing you everything for the channel self-service and it's a little bit different um for sure it has also the itsm hardware troubleshoot here at the beginning based on our demo data but it should show that if it's self-service for example vigilating could make at the beginning there was seven to start with but for sure you can make that for phone you can make that for email or you can make that throw everything into one box and see what automation will give you um it's really I want to encourage you to start with that to find out with automation opportunities you have on your instance where virtual agent or where other things could help and again in our two-week session we will cover clustering which is another technique where you can yeah Deep dive into your data and find out even more automation opportunities with that um just do we have any question that we could answer um no I'll simply um there is nothing left I guess um I will just take over the share or let everybody be aware that we have a next session coming up on March 8th where we there were some questions about um can we use this for HR can we use this this for CSM well you can use automation Discovery for this but we can also use the the underlying technique clustering so we'll we'll handle that March 8th we will show process optimization so how you can use process optimization and what works look at the first ins and outs how to set this up um so I hope to see you all March 8th if there are no questions left then I think we uh we are done for today I hope you enjoyed the session I hope you can it's useful and applying this in your uh your own organization yeah thank you yes thank you Timo we have some helpful tips in the chat from Marcel thank you Marcel and we hope to see you all in two weeks thank you bye-bye