- Hey everyone, hope you're enjoying our Now on Now sessions today. Today we are going to talk about how we improve employees productivity and experience using the AI chat bot. I'm Bharathi Ramadass, Director of AI Solutions and Services. I work primarily with Search Virtual Agent and part of Emerging Technology Group. Today I have with me Dawood. - Hi everyone, I hope you're all having great time at Knowledge. We are super excited to show you how we are leveraging GenAI Now Assist in ServiceNow to improve our employee productivity. My name is Dawood Khan Mohammed, I'm a Staff Product Owner, part of Conversational Interfaces. I work closely with Bharathi. - How many of you know what Now on Now is? picture of hands. We have some. So we are primarily customer zero of our platform products. We implement our products, we work with product to come up with the best practices for customers to deploy the solutions. We also innovate in our own space, which eventually may get into our product. So we are customer zero, which we implement. We use it and also get our value metrics out of our platform products. Today we'll see how employees find answers, what are the tools they use today to find the information they need? We are going to see how GenAI Now Assist has helped improve the productivity. We'll see quick few demos and also wrap it up with a few key takeaways today. Three reasons why we are excited about GenAI at ServiceNow. We think it is a game changer for us. It is embedded in the platform. It's easy to implement and there is no organizational complexity when we roll this out. With our current implementation, we are seeing tangible benefits of 10 million plus internet ecosystem. How many of us are challenged to find the right responses in our organizations? Today we have, we set out regular surveys. We conduct user research studies to find out what are the key challenges our employees find. One of the things we have heard loud and clear is that they can't find the right information. They have too many channels presented, too many applications and various informations available. Some of the key questions they have today are, could vary from if you're a new hire, you want to set up your payroll, you may have to set up your accounts. It could be things like when the next holiday is, no more information about the webinars and sessions we have, or it could be issues with your IT equipment or applications. There are two primary channels. Most of our employees go that ServiceNow, which is search. It's available on the portal. The user will look for what they're searching for. We index various sources bringing the information and provide them the answers they're looking for. They can also click on them and find out more relevant information if they're not satisfied with the responses provided. Search is one primary channel for us. The other channel that we have today is the virtual agent. Virtual agent is where user searches, it starts with the search experience but a few additions or add-ons to it. They'll be able to find topics which could help them self solve or guide them through the process on looking for information or else if they're not happy, they can get connected to a live agent or else they can create a ticket by which means the agents will follow up to ensure the employee's queries are answered. Dawood will now walk us through a typical employee experience. - Now let's look at this from an employee's perspective and how they would go about searching for answers before generative AI and after. Meet Christine, a ServiceNow employee. Christine is facing some issues with her laptop keyboard and she's on our employee portal. Started typing the same like laptop keyboard is not working on the search bar. Christine is gonna get some search results. You could see there are like bunch of articles, KB articles and catalog items. Christine clicked the first one which is relevant to her query that is like keyboard troubleshooting. Now Christine has to scroll through the complete KB article to find the information or the resolution steps to fix the issue. You know what, Christine is not happy with this experience and as Bharathi has mentioned we have another channel. Christine wants to explore that. So now Christine is on employee portal again, but this time she's gonna ask this on the virtual agent. So Christine is gonna type the same thing. Laptop keyboard is not working. But with the virtual agent experience, you could see that instead of KB article and like scrolling through and all that information, Christine has like a better experience of you know getting some steps, a resolution steps to go through it. And the good part of talking to, or conversing with a virtual agent is that, anytime like throughout your conversation if you feel that you are not getting the right response or through all the resolution steps, you feel that it is not helpful, you'll get an option to connect to any support agent. So Christian has gone through all that and she's still not able to fix the issue that she's facing and that's what she is connected to a support agent by clicking onto, you know, live agent support. So we have had a chance to see Christine's experience without Now Assist. Now let's talk about Now Assist. With Now Assist searching for answers can look and feel much different. Now Assist is powered by generative AI and our own LLM, Now LLM for a truly amazing self-service experience. Now Assist uses intelligent conversations to provide employees and upgraded search experience. You know, it improves productivity and you know, delivers speed with a smart recommendations to employees. And the good part of it is, you know, through Now Assist you can also order service catalog through conversation instead of the regular, traditional way of finding the catalog item on the employee portal or the virtual agent providing you some links that you have to click on it and get onto the form. So as we have talked about Now Assist, now we are gonna show you how we have improved Christine's search experience giving her time back in her workday. Now Assist provides summarized response so Christine doesn't have to use multiple clicks and scroll through the complete article. You could see that Christine gets the responses or the answers right on the screen. So this way you know, Christine spends less time to quickly find the answers that she's looking for. So as we have seen how Christine dealt with her laptop issue, now let's shift our focus and see how she plans to order her headset. So let's take a look at the traditional approach. Christine is on the employee portal. She's trying to find the category that where the headset is located. She found that and she clicked on the headset. Now Christine knew what she wants, but again, with this approach she has to click the variables, select the dropdown and you know, the type of thing that they want. With Now Assist forward virtual agent, you could see that Christine is just gonna type with one line query that she wants a Sony headset a pretty long length and you know, USB-C type, you could see that with one query. She got the KB catalog item and it's gonna give them an option to make sure that it's the right one. You could see that, you know, it's highlighted that those three things that they're provided in the question are all slot filled. That means that Christine doesn't have to provide again and the virtual agent is not gonna do the back and forth conversation, and it's gonna give you an option. You know, just let's say you change your mind and you want to change one of the variable. So Christine wants to order a regular USB then type C and that's what she told like I want to make changes. So you see that how easy it is to quickly get the things done. So this interactive and conversational approach is not only going to improve employee productivity, but also gonna give a better enhanced user experience. With Now Assist at ServiceNow, we have seen 67% of employee satisfaction rate and 99% of top a 100 queries that users ask on the search bar. We have seen that, you know, in the previous experience without the Now Assist, she has to click on the KB article and all that. But with our Now Assist we saw that 99% of queries are summarized and provided like a curated response to our employees. So we have shown you Now Assist, but what's next? We are not stopping with Now Assist, we are taking another step on our journey journey. So we want to, you know, improve our employee productivity even further. And now I would like Bharathi to walk us through our future vision. - Thanks Dawood. So we saw two experiences. One was search, the other was the conversation experiences. What if we combine both of this into one, that is our vision or the next steps for us? We want users to engage in a single interface not knowing if search is working behind or the virtual agent is. So Christine, we know she's had issues with laptop. In this experience we're gonna walk you through how VA and search work hand in hand. Christine is looking is reporting a laptop issue. So there are two options. One is that she can either fix the issue, which is a virtual agent conversation. The next is that we know with because it's connected to the items tool that her laptop is up for upgrade. So this is an option provided for her to upgrade the laptop. So Christine makes a choice of upgrading the laptop. Once she makes the choices, there are going to be recommendations on the site, which says when her eligibility for laptop upgrade is you have a KB article there. Christine can make a choice either to go with the provided recommendation of getting the same laptop she has or make a new choice. So she's making a new choice about the next laptop. She wants an Apple or a Lenovo. Recommendations again change as to see how she can pick the right laptop for her work. So as you can see the choices she makes using this interface, there are going to be recommendations provided on the site for her to navigate and make more sense of her choices. She picks up an Apple MacBook Pro and there are videos available for her to go browse the features that's available as part of the laptop, that we want our employees to use Now Assist and come engage with us on the portal. Once she's done making these recommendations, we are going to give her chances to see what are the popular things other employees are looking for. It's knowledge time now. So we have recommendations on key topics that other users are interested in. If Christine doesn't want to go through, there is a pro option for her to engage or start the conversation on other things she's looking for or else she can ask a follow up question on the previous experience she's had. As you can see now, bringing these two powerful interfaces applications together, employees are getting a conversational experience, getting right hands-on recommendations, and also using the tools in the platform to come up with nudges, smart nudges for them. Having said that, we let the few key takeaways we have so far is that with GenAI we can build brilliant and intuitive experiences, which is seamless. We want to practice go ahead with the data and design best practices, which is applicable. So we bring the right relevant information, what Christine is looking for, and we also want to establish KPIs. As Dawood mentioned with Now Assist we saw improvements in come employee per ESAT. We want keep on sending those surveys and measure our KPIs so we ensure that employees get the best experience out of it possible. So having a baseline and as part of our rollouts, having those KPIs actually help. So if you have, want to see how to build these experiences hands-on, you could join us at an hour now Booth on via live hands-on experience where you'll be able to try and build out these experiences to see how simple it is to use Now Assist on our platform. If you have any questions, we're open for questions now. (indistinct audience chatter) Hi, thanks for the presentation. Biggest question is, so if the user has an issue with a keyboard, they might not be able to actually use the desktop experience. So can you talk a bit about how this is integrated with mobile? - Could you please repeat the last part of that? - Yeah. How would this look for a Mobile experience versus-- - Mobile experience? Yeah. - So we are trying to build the same interface on the mobile. It'll be a standalone app is what we are exploring to have that interface built today we have search and virtual agents as two separate applications on mobile, while we build it at the same one, we'll have a mobile application for experience for this. - And also we do have a mobile experience for the virtual agent. So please feel free to stop by at our booth. We can demo you like the current mobile experience. Okay. Okay, thank you. I think one of the biggest questions when it comes to AI and generative AI and especially with Now Assist is, what are the requirements to feed into it to make it successful? How switch on is it? - So today we use our own public data for the training, the models. The LLM has its own curation months. Most of it is done on the platform and one that can be public is what we are sending out to the generative AI. So for the recommendations we use predictive intelligence, which is not generative, but we have it already on the platform. So you get the recommendations here. But for summarization we are using only public data, yeah. So can virtual bot work with Teams? Because employees don't wanna have two, you know, virtual session, virtual bot and Teams implemented sideways. So now that ServiceNow works with Teams, can they be linked together like a click or something? How would you? Do you have plans? Can you do it right now or how can it be implemented? - So the question is, do you have the virtual agent integrated with the Teams? Today we have out of the box store application wherein we have the virtual agent as an application on Teams where you'll get all this information, you'll have a conversation piece, you'll also have notifications that's being enabled on it. It's been, I think for the last, I don't know. - Yeah, so full part of integration with MS Teams or Slack or any other social platforms, you don't need to rebuild any topic or any of the topic flows or development you have done. Whatever experience you have it on the desktop or web experience, is just gonna be replicated on Teams. However, you know, the experience might change a little bit because of the different platforms. So that's gonna be a simple configuration. And the good part with the Teams or Slack, right, you are also gonna get proactive notifications. So proactive notifications to give an example, your 80 password is able to expire in 14 days, right? So this virtual agent on Teams can send you a notification 14 days or you can set it whatever time you know you want to timeframe. It's gonna send you a proactive notification saying that your 80 password is able to expire. Along with the notification, it gives you an option for that conversation. You click on that, I want to reset right now, it's gonna do it like, any topic you want to configure with the proactive notification, you should be able to do that. Hi, how are you? - Hi. I wanted to ask about changing behavior around Q&A So I know with chatbots, a lot of times user behavior is to put in really short phrases. What we're finding is with those GPT solutions that they won't actually generate a response or a meaningful response. And so I'm curious if you had to do any change management around the user behavior to say, "Hey, like you know, if you search, it's not gonna return anything. If you ask a question, you need to give a little bit more context to GPT for it to be meaningful." - So while we started providing recommendations on how to phrase the right questions, so sometimes we have auto suggest working in or we have recommendations like people also ask for which prompts the users to rephrase the user queries that they want to ask. So that's one way that we have implemented or we are trying to find the better way around it. But that's the one that's working effectively for us. Can I ask one more question? Just around access. So I know it will limit what people have access to, but in our situation for like HR, you've got access to global content. But for me, if I'm asking the question about me, I care about US content. If I'm asking about another country, I care about another country's content. How do you prioritize your results based on me as a user? Is that factored into analysis today? - You as a user is factored into it, but for now the global content is one that comes up first. Then the next with whatever content that's relevant and pertinent to you would come up. That's the way it's working. We are improving, trying to improve the ranking system. So the user profile is taken into consideration to provide the most responses, but at this point it's mostly global and public content. So that's what takes precedence. That's helpful. - Thanks everyone. If you have more questions, we can meet us at the booth or we'll be here. Thank you. (audience clapping)