good afternoon thank you for joining us today kolve technology would like to welcome you to our service now Federal Tech talk enhancing Federal operations harnessing Genai with service now's now assist joining us today with service now we have Miguel daer outbound product manager Miguel the floor is all yours good afternoon everybody thank you for coming um if you're familiar with service now and have been to a couple present present ations I have to cover this uh at safe Harvard for looking notice um I am going to show you some things that are coming in the future um so please don't make any P purchasing decisions based on solely what I'm about to show you so without that with that out of the way let's go ahead and take a look at the future of geni and service now so the first thing before we begin the first thing I really want to focus on is service now's approach to this is really around the user experience not just with Gen but the platform as a whole so with that said this is going to be the future for the end user experience and let's take a look right the first thing I want to point out here is we're going to unify that search experience oops we're really want to unify that search experience so we're going to go ahead and combine both the search and the chat capability or the virtual agent capability at the same time so in this in this demo you'll see that the user is having an issue with their meex uh video application but not only do you have the ability to chat with the agent it also has the ability um to uh point out some knowledge articles as well as some other things that are associated with what each haded in there now it's not just a virtual agent bot it's also a way for communication so if you notice in the demo it's explaining what's going on so the the virtual agent is asking for a picture so now we can feed it pictures for it to better triangulate the issue as you can see it's laggy it's not uh pixelated so the virtual agent is quickly triaging the situation and it realiz that it might be a network connectivity so in addition to just talking and chatting with the agent we empowered the virtual agent with actionable items and and as you can see it's going to go ahead and run a Diagnostics on the network to ensure it's not a network connectivity issue came back not a network issue as it's happening as you can see it's also updating the knowledge articles and suggested items on the right hand side um so while it's doing that it also created an incident and alerted an agent in the background what's going on now in a second we're going to take a look at that point of view but for now it's open an incident in addition to the incident it also has the power of the platform so with that said it it has access to personalized The Experience because it has access to the cmdb and anything Associated to that user to include this uh accessories some laptops and the software associated with it so it has insight into what's going on and because of the issue was a video issue it realize that their video camera is outdated so here it is and associated with that it's it's going to allow the user to update their video on the spot so it's a proactive approach and not just a chat approach there it is so it went ahead and ordered it now in the background the ticket's still being worked on during this conversation the agent and the chat and the virtual agent and the agent figured out what's going on and it went ahead and patched it for him it realized it was a software issue it was it was an old software because of the access to the cmdb it went ahead and automatically patched it so again that action will experience the really userfriendly experience here now let's take a look at this same incident from the agent perspective from an agent's point of view it's got to sign a ticket it's VIP automatically let go ahead and look at it so the first thing you notice here not only provides you with information needed it also gives you uh a quick summarization of the incident so on the on the right hand panel it shows summary what's going on in addition it also provides information how to quickly triage the situation what we also introduced is something called analysis panel which is in a sense is like a sidekick for the agent giving it the ability to directly interact with our generative AI solution it's letting that gen know that the knowledge article provided didn't help and if it has any more suggestions and because again it has access to the platform it realized that there's some other incidents that also are similar and their solution was the the software was outdated and again this is a proactive approach it's allowing the user to go ahead and trigger an action to automatically push that update to that user's um laptop but again it's it's more than just a chat feature we really wanted to be proactive again and a approach right so in addition to identifying the issue it also identified there's 15 other users that might have the similar issue because their software is outdated so to prevent those incidents it's allowing that agent to go ahead and push that patches to the rest of the 15 individuals with the issue right now same in that proactive approach it it recognizes there's also a knowledge Gap so it's allowing the the agent to create a knowledge article based on what just happened to prevent things like this happening and to populate the virtual chat that we saw earlier so it went ahead took the information and populated a knowledge article a draft version but not only that we also have the ability to summarize things or shorten things using the power of genad so in this incident you know it seemed a little um lengthy so we shorten it down and now we get to publish it and there you have it that's what the future looks like and it's coming quick you know knock on wood hopefully by the end of the year we should see some of this in in production now you know great intro my name is Miguel daer I'm the outbound product manager for at service now for a platform in geni um I've been at service now little shy of five years I'm about a month and a half short of five years uh but I've been in ecosystem for about 11 years um and my career started in the Fed space so the FED space is near and dear to my heart I'm also a veteran so that you know I really care about the FED space and I'm excited to share this with you and talk talk about geni today let's talk about geni but before we begin we need to understand the strategy and how we place it in the platform and if you're familiar with service now you understand how the platform works we have our workflows that sit on top of the platform things like HR itsm and CSM and they what they do is they leverage either fun ality or products or services from the platform a good example of this is reporting everybody that ever uses service now has some kind of interaction with reporting but reporting is the same throughout the platform regardless if you're using itsm or HR that same experience is the same and that knowledge is the same you don't have to relearn how to create a report in addition to that if you create an application using our creator workflow you still have the ability to create that reporting system so we took that strategy and we start we we wanted to introduce generative AI at the platform level because of that now it does a couple of things when we introduce generative AI on our platform we also introduce another product called the generative AI controller and for a lack of a better word it's really the control center for all things generative a on the platform to include adding those workflows leverage that generi and future things that are coming down the road and by doing this it also removes the complexity right because we're allowing you um to leverage different LM models as well and I'll get into that in a little bit what that means but we're removing that complexity making it easier for you to integrate with these things and Leverage The Power of geni throughout the platform and finally another reason why we added it on the platform was because service now is really good at a lot of things one of them is structured stuff right we have that uh one database model we're really good at structured workflows and automating those workflows but we're not good at is an unstructured data examples of those are knowledge articles right we can't really dig into knowledge articles we can't you know uh pile into that data or search that data same can be said with virtual agent chats and ticket activities where there be a case or an incident but with the power of gen we finally unlock that and we're going to marry the two right creating a massive productivity cataly for our customers now generi is a new tool on the Block everybody's talking about gener AI but our approach is different when we introduce gener on the platform we want to make sure it does three things right there three pillars are really targeting the first one is it has to drive drive productivity for our customers either drive it or enhance it second thing is the value we know that value is important to our customers we know that value has to be at utmost and has to be driven and has to accelerate and lastly and most importantly is the experience the user experience of the adoption rate without that we can't achieve productivity and value so we're really focusing on on the experience and if you notice in the the beginning demo in the videos we really focus on activity and and this is why now on experience the way service now took an approach is by personas we need to accelerate everything people do to accomplish what they do on our platform so with that we took a lens on the personas we know that the developer's Persona in dealing with gener AI and his work or their work is going to be different than the employee and same can be said with the admin versus the agent the customer so we're really focused on the employee and in a little bit in the demo I'm going to show you some of these interactions by these different personas now if you're familiar with service now you've been here a while you know that we love branding our stuff everything is branded it has a name generative AI on service s is no different and we calling it the nalysis uh so it's could be nalysis for HR or nalysis for itsm the analysis is the brand but it does another thing for us as well uh AI responsibility or generative AI responsible is a word that's been tossed around lately so service now has decided to leverage this branding to identify anything on the platform that leverages generative AI by either labeling it with nalysis name or nalysis logo and that's going to have that human in intervention and it's going to prevent certain things from either being published or produced um by allowing you to either delete it modify or keep as is so keep that in mind uh when we see the demo you'll see the the markings and and I Point them out to you now let's go ahead and look at the demo me switch screens real quick all right so the first P I'm going to take you to is through a user experience and oops Yeah the user experience so the way an end user customer interacts with service now is to the now assist for virtual agent on uh any portal and it's it's more of a this in this incidence I'm going to go through a question and answer how to order something because it's something I did now as anybody else I am glued to my phone I like my phone a lot and sometimes I spend a lot of time more than I should unfortunately this morning I had a case of Butterfingers and I broke my phone so I'm going to go to Virtual agent and see what I can do right and it's it recognizes my mistake and it's it gives me an ability to request a new phone so I'm going to go ahead and get started I'm gonna give it a justification I dropped it and broke it now it's giving me options for storage I like taking pictures so I'm going to give it the natural language approach right I'm just going to give it a number and say I want a 256 gigabyte and you understood my intent all right uh I'm gonna go green again understood my intent capture the datam it's going to populate a record for me at the end I want it delivered here okay you know what I really like taking pictures so I'm gonna make a change I need more storage now based on that it automatically knew the next level up of storage was 512 gabyt it didn't need me to tell it again that that the natural language type interaction that we're looking for and all set now so I'm going to go ahead and submit this unfortunately I have no pictures of my phone because I use my phone to take pictures and there it is let's take a look at what that looks like so it requested my item and it's given me an ability to check it and it kicked off the The Rhythm now just look what el v virtual agent can do for an employee um I have been traveling lately and unfortunately I don't really know the hotel policy so I I asked um my manager for the for the policy and he just sent me knowledge and it happens a lot right so sometimes knowledge articles are really beefy and it's hard to find the information so let's see if virtual agent can help me out and there it is it provided me the snippet of the the Huge Non code where all their information's at saving me time right now I don't have to bug my manager anymore and to to make sure that we are getting the right information it's providing me the source where it came from and again remember earlier I talked about The Branding there it is identification of it's powered by n assist now let's take a look at the article to ensure it came from the right spot as you can see within this article is the heel policy now let's go ahead and take a look at an agent's point of view of this earlier we saw the agent's interaction so today I got assigned a ticket and if if anybody Works a ticket or has worked a ticket they know how complex it can get especially here in the activity section because now only you have to read the back and forth between other agents and the customer you has a track in your head what Solutions worked and what didn't and where they're at the last time I did this it took me about an hour just to figure out where they're at before I can even engage in the ticket now what we had introduced here is called instant summary and what this does is it goes through that Combs through that and provides a meaningful summarization of what's going on for me to take action and there it is it's letting me know the issue and the key action is taken okay now a second pain point in working tickets is the resolution notes as oh and I can sorry and I can also share this so here we go here's that section where I talked about the human intervention where I can edit this to make sure it's it's justifiable and then I can save it to work Nets so there's there's not really a point where gener of AI can just public things for you okay now second pain point is resolution notes uh I just it's it's painstaking to go through it again find resolution notes so what we did was we decided to go ahead and pre-publish those for you create a draft when you hit the resolution and there it is brought resolution notes looks good to me I still have the power to edit if you need to so letting me know it's provide of analysist and I'm going to go ahead and solution provided resolve the issue and there it is something that would take a couple of hours to figure out to populate or to read under understand took a couple seconds now earlier in the video we showed analysis of the sidekick I called it and here it is here it is this is analysis panel today and it does more than just chats back and forth it can also be actionable I know there's a knowledge Gap because the tiet said there was no knowledge article so I want to go ahead and request that it generates a knowledge article for me there it is it created a draft article based on the information in the ticket and now I have the ability to open that ticket or open that knowledge article and again edit the draft as needed before it gets published that's the power that's where we're driving right efficiency the user experience right and as you saw what the feature looks like and that's where we're at today last not least let's go ahead and take the view of a developer what we have today is something called text to code and what it does is takes simple simple text commands or requests and it makes them into code using service Nows now llm and I'm talk to you I'm G to mention what that means later on in this instance I need to retrieve as a low code user or as as somebody that's not service now code I can ask it to to retrieve all incidents created since the beginning of the last year now if we review the code before we accept it we can review the code ensure that's the right code for what we asked for it looks pretty good to me and this is pretty cool it uh found this method um to go ahead and figure that out instead of creating a new script include to pull that data so again it's trying to Seine that approach for our users and if you notice analysis logo is here letting you know that it's powered by analysis now let's take another look at a developer approach let's look at the flow designer view we know that sometimes flow designers can be a little tricky or difficult to use especially for a new user uh but as a manager of an incident I realize I have an issue that not all incidents are getting created or getting assigned to within the 24-h hour so how do I fix this on my own well with now assist I can create a flow to address that so I need to assign all on assign incidents in the last 24 hours now I want to tell it what to do I want to describe my actions to it and all I'm asking is to create a flow that runs every day at midnight then find all the new Crea incident records for that past day if they're not assigned I sign it to the incident one triage group then send notification to the group not complex it's a pretty common use case let's go ahead and build up now assist and as you can see I created a flow triggers every day at midnight it looks up the records and it creates conditions based on on the need now we're empowering more people now that are scared or fearful to use flow designer we're empowering them to use it more with using analysis and generative AI all right let's go back to our presentation all right we saw a couple of glimpses of what gener AI analysis can do and it's great but we're missing the bigger picture what does that mean for you and the value it's good to drive um it's just it's not a clear picture and I understand so let's go ahead and take a look at how that value can be driven for you here's the timeline of an incident starting with the employee trying to find a knowledge article sorry to solve to solve the self sorry here's an employee trying to find a knowledge article to help help self-service the issue unfortunately no knowledge article exists so the employees natural steps to go to Virtual agent after a few attempts the employee decides now it's time to speak to a live agent the first thing the agent needs to do is TR a conversation so that no information is missed and then no answer or no question is answered again understanding the issue the agent then tries to see if a knowledge article exists unfortunately no knowledge article exists fortunately for the agent agent assists finds similar incidents however the resolution notes on the incident hasn't been properly populated by the previous agent not knowing what to do next the agent has no choice but to escalate the incident to a tier two agent the cycle repeats again the tier two needs to understand what's going on in the ticket understand the actions taken up to this point in order to know what next steps to be taken are after resolving the incident the agent is asked to populate me for information through resolution notes unfortunately the agent doesn't have time for it the same can be said about creating a knowledge article the only thing the agent has time for is to a report a knowledge Gap now let's take a look at that same process powered by the analysist the first thing we notice is the chat summarization creates a summary of the conversation between the employee and the virtual agent saving valuable times to the agent when it has to triage a ticket the same can be applied when the incidents hand over to the tier 2 agent where the instant summarization provides a quick and concise summary of the activity stream saving the tier 2 agent again valuable time resolution notes are not automatically created providing for best practice and it'll help the next agent that runs into a similar incident same for creating a knowledge article that helps the agent adhere to best practices and it helps the next agent again find a quicker solution moreover and this is the key part the knowledge article will now help the Employee Self-Service the issue preventing the creation of that incident and last but not least the knowledge article will also show be shown in the AI search result and the AI summary the AI Search summary that will provide a better experience when engaging a virtual agent if you look at the kpis that now still self improved right we're driving that value the first one is the employee side where it only Prov sorry where not only proves better information to to drive self-service it also proves that provides that better user experience uh both the chat and instant summarization helps reduce that meantime the resolution while creating while the creation of resolution notes and knowledge articles Save times for the agent and help supports best practice while at the same time the improved knowledge base and resolution notes help reduce the meantime the resolution increase that firsttime fix for incidents and allows for few escalations and this is how now cyst for a single process both improves the user experience and the agent efficiency but this doesn't apply just to itsm or an incident this can be this is true across the board for all our workflows and here are some example values that can be driven okay I know it's a lot so I'll take a pause real quick here say if any questions out there any answers or something that came up there's a question that says out of curiosity is this assist AI technology a default feature in service now so I guess uh so we can go back up so the nist is isn't a default feature yet you do have the right to select what skills we call them shows up and doesn't show up so if the nist panel that that little assist panel the psychic if you don't want it to show up you can definitely have control about what shows up and what shows up same thing with the summarization you have you can uh you can determine um who can view it and where to show it so you do have some control how to manage those things on platform I hope that answered the question any more questions not seeing anymore awesome now let's talk about what powers uh generative Ai and now assist um earlier I mentioned the AI controller or generative AI controller that was sitting or put in a platform uh because of that strategy we're allowing for a flexible more inoperable Ro when it comes to powering the generative AI in the system on the platform um and if you're familiar with large language models um we're going with a two prong approach one general purpose which this is uh they third party systems that you can either buy license for bring with you um and currently today in the FED space we have it set up for Microsoft Azure now the general purpose are trained in about 80% of the internet they're really good at that natural interaction that chat interaction as well as creating content um you know I think my friends create their their papers on that right and some some blog post so it's really good at that um we also are introducing or service now actually invested a lot of money into our domain specific um llm and we call that the now llm and what we did with that is we know service now better than service now so we took all those 20 plus years of of best practices coding process power data and we build our own domain specific element cater to service now and and a good example of why we went down this path there's two good examples the first one I'm going to go over is the coding aspect I showed you the text decode earlier demo of that and we uh actually took uh that same question not that one but another coding question and we presented it to the general purpose open AI model and It produced us it gave us a result and it was a really good result like the code was really structured it was really commented out however to the untrained eye there was a couple of methods in there that looked like they belonged to service now but they didn't exist in service now uh so you can see where where that can be leading that Snowball Effect down the road uh we took that same code and we put it in our domain specific llm and It produced again really optimal code really structured code really nice code but more importantly we asked one of our developers to write the code itself and that's like our base method and we realized that our domain specific LM not only wrote the code it wrote the optimal code um if you're familiar with sarus now you you've been here a while we know that we're constantly updating methods and providing new ways to query our database and code and sometimes we leave some of that old stuff behind because people are still using it our clients just they're using it but with this code it went for the offal rout right off the bat so again it shows you the power of what it can do in addition to just knowing what we know one of the biggest things that came out when generative AI talk first started was security um of course security has always been our top priority and with an nlm we were able we decided to go on that path is because of that security because we own our own llm we maintain it we care for it it also resides in our data center because of that it it has to it has to apply to the same security standards and policy we have in place today and we don't expose it to third party vendors or third party systems nobody can access it except for the service non instance we also give you the option to op out training instance with your data and because we're talking in the FED space the gcc's l m or now llm will never interact with a commercial now llm it could be completely separate again it is still apply by the same security policy and constraints there are today um however the general purpose um llms we can't guarantee any security after it leaves our boundaries so while the data that resides in service now is there and we don't leak it out you have the choice to actually submit that data to their llm and once that's transmitted we have no clue we have no Insight on what happens to the llm I get a lot of questions in this aspect so are there any questions on this so far no can the AI model be trained specifically for the organization do we as the end users have the ability to refine that level of training without a service now technician or assistance if we have that level of knowledge uh not not today um it is something that we're looking at and I seen some things um at some of the at some of the meetings we have but not today everybody wants to talk about the road map especially in the FED space the second question I get the most outside security is when is it coming to the GCC so happy to announced it's coming in June of this year the summer of this year we will be supporting the now llm in GCC the National Security Cloud the NSC Cloud currently the demands under review and we don't really have a date yet um once we decide on a date or if it's going to happen we'll obviously let everybody know but it is on the review and last but not least the self-hosted arm Prem solution self-hosted arm Prem is near and dear to my heart I spent a lot of time in that space uh So currently today as we speak there is a pilot program going on um we call the go to market phase we are taking that feedback and enhancement and we're making changes and there will be a limited roll out in either Q2 or Q3 now of this year now I'm going to say a couple things on that the now Alum there's some certain criteria you have to meet before you can um get your hands on it um there is things behind it so if you're interested again I would suggest you reach out to your account reps uh they can help you see if you meet those criterias and you know get you along the way one more thing I want to talk about the now for self-hosted on Prem now llm solution is it requires an investment on your behalf for the the infrastructure of the housing because the now llm is massive it requires a lot of processing power we can't just stick in on an inst an instance right we have to give it its own infrastructure so just keep that in mind if you decide to go this path those are some of the issues and requirements that come along with this solution again if you're interested please reach out to your account reps they have a team that supports them and they'll know they'll get you on the path to to using one of these Solutions right now let's go ahead talk about the road map for the now assist now we we we went ahead and showcased you some of the stuff that's coming out um I demoed the multi-turn conversation you saw me order an iPhone the knowledge creation uh it's really great I think I had I recently talked to a customer a big Energy company and he was telling me he had like three something knowledge articles that are on backlog and when he saw that he was excited because now instead of putting in a backlog the the the agents can create it and and the knowledge team can go ahead and take over so it's really excited I think it's a really powerful move the fla automation again I just showed you how you can take text and create flows all that is coming within the next month or so um another thing I want to talk about is that we are doing uh four releases a year uh we're trying to release them to the store app to get these products to you quicker and you can get them down to your production instances faster we do have certain things that are coming out within our general releases the two releases a year but we're also doing store plac now let's talk about what's going on uh the video we showed you earlier in the beginning uh touched on some of these so you kind of have an idea what these mean but again Q3 around the Washington xand do time frame a little bit after Washington we're looking at introducing that seamless service now a self-service experience with the chat and the search capabilities uh reply generation uh it it was on the video but I forgot to call it out but what this does is generative AI can create um responses to either chat or email which is pretty nifty I saw a demo of it it was in a demo forgive me uh but it creates those email templates for you or email responses for you uh and then finally in the creative side we're allowing you to bring your own llm uh this is the second the third ask I get U when I talk to clients is can I bring my own llm we are going to start introducing that with the controller around quarter three and then lastly for the end of the year round and out the end of the year we're looking at introducing the the personalization Conversation Piece I mentioned that earlier in the video where the the now assist knows what assets you have and capabilities we're also G to allow it to know where you're at location wise so if you asking something that's only catered to North America it's going to know that uh same with Europe proactive help we saw an introduction there where it's ask the generative AI or analysis for help had any suggestions that's also going to be introduced in around quarter 4 excuse me and then finally the global assist that that now this panel uh we're going to introduce it across the platform as well not just the agent view we're going to introduce that to everybody to use all right I think my I think my time has come to an end so I will open it up for questions and answers if anything's out there if nothing else uh thank you again for letting me come in and uh spend time with you um talking about gener Genera of AI and Analysis I'm I'm really excited about this um I hope you guys gained some knowledge out of this and thank you again for letting me come and talk to you any if you're interested in information obviously reach out to your account reps um we're we're getting up to speed with documentation and getting the community out there um the so when it's there I obviously send the link and if you need any more help any more assistance and you want me to come in talk to any clients I I I can't promise it but I will try to make my best effort to speak to the FED space so if you're interested let your account rep know and I'm more than happy to come in and talk to some of you guys thank you team I'd like to thank all our participants as well as our speaker for being with us today we hope the information you received during this webinar has been helpful if you have any further questions or would like to request more information please feel free to reach out thank you again and have a great day