The role of AI in self-service and knowledge management — Kajabi's Jared Loman and Zendesk's Maddie Hoffman

From the get go our philosophy has always been how do we make the customer experience better. That was the one thing that we had to hold true. Like this cannot and of course if we're thinking about metrics like we cannot let CESAT suffer as a result of us having this as a weight in the support flow of one point in the journey. Hello and welcome to the conversations with Vendes podcast. Each week we speak to customer experience innovators and experts to discuss the latest trends, interesting ideas and what's new in our industry. I'm your host Nicole Sanders. Last week we dove been talking about the business impacts of AI. We spoke with Jared Lohan VP of customer experience at Kajabi as well as Caitlin Corhane who is the head of customer advocacy here at Zendes. They shared some great insights on how AI might play into the aging experience and how you can save money as you scale your customer support. This week we're going to focus a little bit more on self-service support in the role that AI made slated there. So as I said Jared is joining us again and Jared is an XCTO and engineer turned CX enthusiast. His tangible expertise combined with his passion for delivering exceptional customers experience has earned him a spot on several advisory boards including the customer advisory board at Fort Thought AI which is a company focused on delivering artificial intelligence solutions for customer service organizations. During his four years at Kajabi as the VP of customer experience he has challenged his teams to be both early adopters and innovators of new ways to leverage technology to empower at agents and provide better customer experiences. Today we're also joined by my friend Maddie Hoffman, director of self-service support at Zendes. Maddie has deep experience in customer support building and scaling self-service support chatbots and automations. In the seven and a half years that she's been at Zendes she has helped to build out the way that we capture and share knowledge in our health center. So let's dive into my conversation with Jared and Maddie. Reddies, take your customer experiences to the next level. Build lasting relationships, the Zendes complete customer service solution so that you can exceed every customer's expectations. Sign up for a free trial at Zendesk.com. Hi everyone, welcome to Conversations with Zendesk. I'm your host Nicole Saunders, director of Community at Zendesk. Today we have our second in a two-part series on how AI can help to rain in cost while maximizing efficiency. I'm joined today by Jared Lomon VP of customer experience at Kajabi and Maddie Hoffman, director of self-service and automation at Zendesk. So excited for this conversation. Thanks for joining us Maddie and welcome back Jared. Thanks for having me back. I don't think I've ever been invited back so this is a first, I appreciate it. We're glad to have you again. Before we dive into the conversation, I would love to have you go ahead and introduce yourself again and just let us know. Tell us a little bit about your role, your organization and what you do. So Jared, I'll have you start. Yeah, so I'm Jared Lomon. I oversee our customer team at Kajabi, which includes our technical support, our customer success, our knowledge and training, the small ops arm. Ben with the company for about four years and we provide a tool that really plays a big role in the creator economy space that helps anyone who is interested in selling their knowledge online and I am calling in from the beautiful state of Hawaii. Oh, jealous there. I'm up in Wisconsin. It's just starting to get warm here. So warm 40 high forties or like actual warm. We hit 70 today and that was super exciting. All right, Maddie, please introduce yourself. Absolutely. I was going to say I did have to take my jacket off on my lunchtime walk. I got a late to warm. I haven't felt that way in a long time. You're in Wisconsin. Hi, I'm Maddie Kaufman. As mentioned, my team at Zendask is responsible for self-service access to knowledge about how to use our products, how to get started, how to configure, how to test and validate, as well as automation and our virtual assistant or chatbot that's at the front of all of our interactions. So very focused on getting resolutions and answers into the hands of our customers faster and automating their solutions wherever possible so they don't even need a human's help of it. All right, as we dive into the conversation to recap in our previous episode, we talked with Kate Cahane, head of customer advocacy here at Zendask in Jared. We spoke with you as well about how AI and bots and that kind of thing in general may start to influence customer service operations, create better customer experiences and that kind of thing. Today, we're going to dig in a little bit more specifically into knowledge management and self-service. Now, of course, where I think most people are really familiar with AI right now is in terms of bot experiences. And so I'm going to start there and then we will move to more broad knowledge management. Earlier this year, we did some research and it showed that consumers really think that AI is going to lead to much more satisfying chatbot experiences, which is great news. So I'd love to hear from each of you how you think that AI will start to improve those bot experiences and how we would get to a point where those bots aren't just conversationaly sound but also really accurate and truly helpful. Do you like me to go first or should you give Mattia a chance? I'm happy to dive in for sure. All right, I think the age that we're entering really is going to be like a totally new horizon of how we find information. And I think when I look at things like LLNs entering the scene, for me, it's less about generative output, a bot designing an answer for you and more about this totally new web of connections that can get you from your question to an answer by connecting the dots. And I guess by that, I just mean we've already been a little spoiled in the world of Google that we have the most powerful search algorithm in the world working for us most of the time. But I think all of our end users are going to only continue to have heightened expectations of how easy it is to find information. And in my opinion, that's where I really think AI introduced into a conversational environment like a chatbot makes a huge difference. If your customer's not going to be frustrated by feeling misunderstood or wind up in like a dead end or a loop because something that you historically might have had to train to be really maybe if then it can blossom into something much more conversational. And I think there's a lot of really interesting chatter in the industry right now too about taking some of the problems that we're used to solving in say for example, I work in software so in an interface and a product and shifting that to be solved conversational, solve problems conversational with our coworkers every day on tools like Slack or teams or messaging one another. Hey, can you do this for me? Can you do that for me in seeing action happen? And I think that's going to pivot the expectations that our customers have as well. And when I think about the quality of the conversation, I think obviously the hot topic of generative AI has been like on the front of everyone's mind for the last several months. I really think the secret sauce to quality of output is keeping a human involved. I realize that some of us may have a daydream of being able to start using like an AI technology where you can just flip a switch and it works right away. There are certainly like plenty of tools emerging that make it really easy and fast to deploy, but I really think if you want to guarantee the quality of your interactions, you resource it. You assign someone to review the outputs, understand what conversations look like, understand what your customers needs are at various phases of their relationship with you, and really have like ownership over that, adopting the customer perspective and then validating that it's working and that they're not getting stuck in some of those more historically frustrating chatbot conversations that so many of us are accustomed to. Absolutely. I think the support agents of the world are probably relieved to hear Maddie that you feel like humans still need to be involved and I totally agree. It's closed out. The last podcast talking about the value of the human touch, even as we evolve these more AI driven experiences, I think that the human involvement is actually going to be more important, not less. Jared, what are your thoughts on this? Yeah. Just to play off of that, I really like to think about the technology again as a superpower. It's a way to power up the human that's involved with whatever the task is. And so I'm less maybe afraid of an apocalyptic situation if you will to where all of the jobs fade away. What I think we'll see is, was you're going to see a transition and I'm actually pretty optimistic right now that in terms of the emotional interaction that we have, I'm excited about AI's ability to adapt, especially with things like sentiment analysis. In a marginal world to where AI comes in and first, it maybe starts with the de facto brand voice of ultra empathetic, but then maybe transitions based upon the conversation into something a little bit more deliberate. And instead of I think maybe short term next year and a half will get to a point to where the AI can maybe even transition to automatically just escalating right away. I can see your frustrated. Let's get you a human. Let's not even waste time making any angrier than you already are. That's good. I'm glad to hear it. Thinking about that idea of a bot adapting to a conversation or to your boy, they're like, "Whoa, your frustrated. Let's escalate that one." How do you think we can best go about creating bots that have a high degree of empathy and are able to be really accurate, but not to the point where it becomes uncanny and too close to human? I feel like that's a needle we have to thread. Do you either of you have thoughts on what that evolution for bots may look like? look like. Yeah, I can definitely say to some of our learnings at Zendesk. I think first and foremost, like Jared hinted at, I think being able to understand where there might be topics that your customers ask about or issues that your users face that are just not necessarily appropriate for an AI to handle. If it's going to be too complex or if it's something that might be psychologically sensitive, I know we see a lot in our world that customers who have questions about their billing or invoice may just really want the extra validation of the human to tell them everything's okay. You're not going to run into any problems. I think it's important to just recognize what those topics might be and as we develop these capabilities, potentially pivot a conversation to invite a human to join in. If it's going to be more sensitive or if we detect that sentiment is trending in a really negative direction and the other piece of practical advice I would give that we've done at Zendesk is define your style guide really early, especially if you're going to have more than one writer building conversation design into your bot. We tried to lay out some really clear boundaries like our bot will never pretend to be human. Our bot will always acknowledge that it is robotic even in its tone at times and even really specific rules like our bot wants to be helpful but should never be overly enthusiastic. We really want to avoid too many exclamation points. Our bot doesn't use the i pronoun. It only uses a wheat pronoun because we want to establish that bot is an extension of our brand of our corporate brand and not trying to make it an entity in and have itself. That's a theme we really try to reinforce throughout conversations as well as that this bot was curated by people who know your experience and are building this so that you can have it at scale but that it's not just pulling answers out of thin air but rather the topics that our customers tend to ask our bot about are somewhat repetitive and it's pretty common that a customer would have had the same question before. I think that can instill a little bit of extra trust. And I know sorry a long winded answer a theme that came up in the last session as well that I would just add on is I really think the more we can personalize the answers our bots can give the more empathetic the conversation becomes as Kate mentioned in the last session. We know we've heard from our customers that they can really lose track of their entitlements what comes with their plan and what features are maybe even with their role that they have using Zundask and our bot is actually trained and has access to some of those data points so that it can answer with practical guidance rather than just like a theoretical which I think mimics a human in the sense that you're getting a targeted answer but doesn't necessarily try to present that as though we're a human I think that kind of like fast logical resolution can be justice satisfying for customers without having to tell the line as you said into the uncanny valley. Those are really interesting points especially about things like style guide and tone and how does the bot represent itself and present itself to the end user. Jared what are some other ways that Kajabi is thinking about how you're implementing your bots and what their personalities might be. Sure yeah I think from the get go our philosophy has always been how do we make the customer experience better that was the one thing that we had to hold true like this cannot and of course if we're thinking about metrics like we cannot let C sat suffer as a result of us having this as a weight in the support flow at one point in the journey. So our goal was always to provide customers answers faster it's essentially the same strategy that you'd apply to self service in general like you're trying to find a way to give customers the answers to their questions as we all know and it was also I think mentioned maybe in the previous years Zendesk report the preference stack which was in like the high 60s so don't quote me on this one look at the report but it was really high for people who actually just want to find a self-service answer and our philosophy with the bot is just it's just another enhancement to that self-service flow it's another way to expose information to customers and help them find stuff that maybe they wouldn't found through a different channel. I think going back to the personality unfortunately like our approach to this is always been like we've utilized our own macros that we pre-vet for like answers to questions so that kind of just inherently carries our personality since we're programming it however I think we're on the verge with large language models of potentially being able to actually really outside of just like creating scripts like we're actually going to be able to train the technology to speak in our language probably better than any of our human beings could take on that brand personality so I would say stay tuned for the next year and a half should be really interesting in terms of carrying forward I think chat GPT for example has done a wonderful job with this like the adding that morals and the ethics component like it will not answer questions that are even like potentially sensitive and so like we're seeing a lot of really cool new opportunities to I think in the very near term begin to actually train this into where the machine is actually able to express this without us having to do all of the work. Absolutely so one of the things that we talked about in the previous episode was how AI is more than just bots right that tends to be where the focus is because it's really exciting they get to go talk to a machine that seems like it can talk back to you and seems like it's thinking. Maddie I would love to hear what are some of the other applications of AI that you're looking at across knowledge management and self-service? I hinted at this earlier but I think the biggest opportunity to find crossover between what we're doing with things like virtual assistants and what we're doing more traditionally in knowledge management is thinking about how you service information to your customers much like you were saying Jared to us as well I think of a chatbot as another tool through which you can find answers. So we've been trying to think a lot about a holistic like end-to-end user experience whether I enter through the chatbot or I enter through a Google search to your knowledge base and land there how can I choose my own adventure to get to an answer some folks actually would prefer to pivot to talk to the chatbot even if they find an article first or in some cases we want them to be able to launch an in-product tutorial if it's something that the bot would otherwise be walking them through step-by-step with text. So we're trying to sort of the boundaries between what we can do with our tools in a way that's a little more autonomous for the customer and along those lines another thought we've been exploring is how we could augment the search mechanism in our knowledge base to maybe even match the intent detection model that our bot has today we really love what we've seen as far as the quality of that intent intent detection goes and frankly I think we're training our customers who use the bot to search in a specific way we're training them that they don't have to have all the right phrasing they don't have to know the jargon they maybe don't even have to speak English as a first language and they could still get to an answer that's relevant to them and I think it's really going to push us to think about some of the ways we surface information maybe in a more traditional approach in our knowledge base and align those experiences to be more in tune with one another and then the other thing I would just add is that we've been really inspired by some of the no and low code AI automation tooling that's coming out in the industry and particularly those end us partner who we work with to provide our virtual assistant today has some really great deep integration capabilities so we're really focused in this first half of the year on where we can actually like completely end-to-end automate tickets that are more transactional today that our teams don't really need to handle maybe it's pure objective decision making that a bot could do instead of a human or where could we save time we're starting to do a little research into conversations with our support team that require a lot of back and force today and how could we maybe get more of that in with the bot conversation before the bot hands off to an agent as well in a way that is of course not frustrating to customers we don't want to increase any effort on their part but if it makes the resolution come faster in the long run then we think it would be more delightful to the customer as well there's a lot to unpack there but Jared I want to give you your chance to weigh in and understand could job these thoughts around self-service knowledge management and how AI might thread through all of that or help to improve those experiences yeah I mean it's going to thread through just about everywhere but I think one of the things that I would call out that is I've seen exciting a lot but it does excite me as gap detection I think that is an area that nobody has ever woke up and said I'm really excited about gap detection but that's what makes it so great for a computer to actually resolve these problems we here's our wonderful handling repetitive tasks that actually humans are probably not the best at and that's a great example of something that I think AI could do really well is identify those patterns and find out what are we missing that no human is going to sit there and review thousands upon thousands if not millions of interactions to ultimately find and identify and in particular with the large language models that's almost getting like this super boost on top of what the technology was already capable of when it's able to really understand the context and do this so much better than it ever was before so that's probably what I'm most excited about of course I'd be one of the things that I just teen off of what Maddie said I'm excited about is language I think that's huge that's something that we haven't talked about a lot and I don't hear about a lot maybe I'm just not in the right conversations but like a bot can speak essentially unlimited number of languages whatever you can program it to do which is not something that you can do with a human like as much as I want to try to learn other languages I'm barely mastered in English. We're all looking for one really good maybe two. Exactly. That's another one of the superpowers on steroids that could potentially change the game for you just about overnight. Absolutely. I find the mention of gap detection really fascinating. I've been reading a lot about the ways that this can play into things like helping identify where do you need articles written, where do you need things updated, where are there things that agents are answering in tickets that aren't documented in your knowledge base. And Maddie, I know that you've worked a lot with how to do this from a human perspective in the past. So that seems like a place where we could really gain some great efficiencies and help build out those knowledge bases that are really solid. And Jared, the point about I've interviewed hundreds or thousands of articles, bots don't fatigue the way human might, right? The 30th article I'm proofreading, I might start missing something here there, get a little bit less focused, but a bot can sit there and proofread every darn article in your knowledge base and make sure that it's all correct and correct. Bringing up your humans to do other more complex tasks. Yeah, I was just going to say you made another light bulb go off for me as well, which is yeah, I've been seeing some really cool applications of just thinking about the things that fatigue a knowledge worker of any kind, including our agents ticket after ticket. I think another really cool application that's been emerging lately is suggesting action to your agents. If you're picking up a ticket that you've never handled before and you don't know where to start, having recommendations that are based on a trained model. Here's how we think this kind of the ticket should typically go. Or I've even seen some technologies out there like this conversation could be escalating. You may want to involve your manager or it could be an upsell opportunity like all kinds of things that we could really prompt our teams or arm them with information. So I think thinking about the previous episode about controlling costs, I think about being a new agent and having a little bit of AI assisted guidance on how to handle something I've never handled before. I feel like that could really have potential in the future to reduce the time it takes us to get our teams live and comfortable responding to tickets as well. That's a great call out. You can really reduce your time to value because you don't need to spend as much time in training if you've got an AI that can help train along the way. And gosh, that's got to be a lot faster than walking down the hall trying to find somebody that knows how to handle a certain kind of situation or pinging a big group chat message and saying, okay, anybody know what I do with this kind of circumstance? Yeah, I hadn't even thought about those. Those are great examples. So where I was going with that line of thought, Maddie, I know that you've been building out your team particularly around automation and chat about over the last couple years. How have you seen the introduction of AI start to evolve the roles and the capabilities that you need on your team? And I'm interested to hear from both of you, what you think some of the roles of the future might be in knowledge management and self-service support? Yeah, I love that question so much because I think there's so many different ways you can do it. So many different ways you can think about this. I've seen corners of the chatbot industry where people tend to focus on conversation design alone, content design. I've seen corners where people tend to focus on engineering skills and how do you build out the technology and just speaking on behalf of my team, we've really taken a hybrid approach where we tend to work with like low and no-code tools wherever possible, of course, but really try to make the most of my team skill sets of both knowing our technology, knowing what the capabilities are, especially as you've mentioned Jared, like in a world where technology evolves, it feels every five minutes, there's like a new capability that we've unlocked. But also, I think just as importantly, their deep knowledge of our customer. And I really think it's an area of business self-service really overall is that requires you to be able to put yourself in a customer's shoes over and over and over again. And I think all these automation technology, virtual assistants and AI, I think it's a really fun, challenging space to be because you get to play both parts, right? Like you really have to adopt the customer's point of view, but you also get to think like an agent and think like, how would I give an answer to this question if I was just having a conversation with someone one on one and from there, how can we expand that at scale to be resolvable for anyone who might come and ask that question? I think it's really taking a lot of the skills we've had for years in knowledge management and blowing them up to think about how we make with theoretical more practical and how we take something more general and make it more specific and just so really interesting convergence of kind of all of those skills. Can't on yourself, of course, to the importance of content capabilities being able to write or produce in whatever media you use, I think that background has been really important to my team because the knowledge of your customer's journey before the ticket is what we always say. The everything that happens to the left and the timeline, I think is really important because that's really where you're introducing the technology at first and from there expands into other parts of their experience of their journey with your company. But Jared would love to hear what you have to say to you about your team. Just speaking, I think more broadly, over the next one to three years, I believe we're going to see a shift to more QA focus. As we become more and more reliant on technology that is maybe not in its, let's never going to be fully mature, but it's still not yet at a point where the state is completely 100% reliable. We're going to involve humans with the QA process to take that baseline that's being presented to us and maybe layer on some of the things that we've already talked about. Here's our brand voice on top of maybe this article that we used GPT or some other technology to lay us a foundation. In addition to that, my hope, this is maybe more of a hope than it is a prediction, but I think once we make it past that phase, my hope is that we'll actually see more people invest or reinvest in richer forms of communication. One example of this could be is like, now we have so much freedom that instead of interacting over a live chat session to where I'm dealing with three of them at once, I'm not just hopping on a Zoom call with you and we're looking at a situation like real time face to face, like really having that high touch engagement that you maybe previously could only expect if you were some top tier enterprise customer, we're now essentially democratizing this to everyone because of what technology is enabled us to do in terms of the volume. >> Yeah, and Jared, I would add on to that. I know one thing that really gets me excited is thinking about how can we layer on to some of those traditional thoughts about segmentation of our customer basis, what we know to be true about certain kinds of issues. I can think of examples that we want to build out as automation that are going to be best served as automation, how our customer work with whatever our automation tool is to resolve it because that's going to be the fastest even if you are that top tier customer. Maybe that kind of issue gets resolved on that channel as opposed to no matter what kind of customer you are, to give an example, if you're setting something up for the first time, you're close to go live and you're just stuck on one part. No matter who you are, we'd love to have you pick up the phone and be able to just walk us through what you're trying to do. I think we can get a lot more innovative and intelligent in that space around not just our segmenting our customer basis, we know it, but our issue base, the kinds of things that we are able to tackle with self-service and automation versus the things that should always have a human on the right channel at the right time. There's some great points. Jared, I asked you last week about what you were most concerned about, what you're most excited about, and you had lots of things you're excited about, which is awesome to hear. This week, I want to hear from you about what has been the most unexpected or surprising thing that you have seen come out of AI when you think about a business application, what has come up free as maybe like, oh, I didn't even realize that or oh, that's a brilliant way that we could potentially use this. Yeah. So the most surprising thing to me was like, when I started this journey, again, I think I said this last week, but I came into this with rainbow glasses and I'm like, you asked the magic technology that is going to solve all of my problems and then I realized that the technology was actually some limiting in terms of my ability to deploy it in the ways that I wanted to deploy it. I think in other surprises, I guess most recently, I wouldn't have said that I would have predicted the amount of like, I'm sure many people have heard of the term generative AI. Like, I didn't necessarily see that coming. I wish I would have. I guess I don't know what advantage it would have given me, but I think that exposes a ton of new opportunities. I touched on this briefly in the last episode and I'll maybe cover the inverse of that. Like, what of a machine is able to actually generate what like a screen recording of how to do something at some point right now. We're still in the point of changing someone on a motorcycle to be in a grill on a motorcycle, but I think very realistically, like a machine could actually generate like a full-on tutorial in real time of how to accomplish a task. And that's something that I think, at least I think, this is completely unheard of. Nobody's talking about this. Nobody's doing this today, but something that generative AI is actually going to be highly capable of doing, perhaps in the fairly near future. So generative AI was my biggest surprise. Very cool. What about you, Maddie? Anything catch you off guard? I definitely can say I relate. Jared, I heard actually recently about someone using asking chat GPT to write code for them that they were going to use to deploy a Zendesk feature event. And I was like, wow, I just never in a million years really would have fired those phrases like strung together. So that has been definitely swept me off my feet a little bit. But I really think it's been most interesting to see some of the democratization of AI tools and really the birth of some of these no and low code opportunities. Being relatively new part of the industry, it's just been really interesting to speak with other folks who every time they want to make a change to what their chatbot says it requires a pull request from their engineering team. And I think we're really living in a whole new world when like my team can go click some buttons and have something done in a matter of an hour. It opens so many doors. And I know we touched on this already too. Thinking back to last week's episode and talking about controlling costs, the things that organizations have gone through for years to localize their content or to hire talent in areas where they need to be able to support a localized, which I think really we're opening a ton of doors to also think about how we can, I don't know, maybe just re-stratugize what we think about localization as organizations. I think there's going to be so many new opportunities. It plus my mind a little bit that there can be models that I don't have to train that I can understand what I've already trained in English in the 100 and some other languages. That's a really crazy thing for me to think about. Really though, I think as we've discussed in this corner of the industry, your mind gets blown a little bit like every five days or something like that. So I'll probably have a different answer next week. I was going to say I can't wait to play this back in six months and be like, wow, we had no idea that this was going to come or maybe some of these predictions will come to bananas. Yeah, I'd just quickly layer on what Maddie said was talking about democratization. This is a little bit of a throwback to a conversation I had with Zendeska a year or two ago when we were talking about the pandemics impact on growth. One of the key indicators to me was like, my grandma now knows what Zoom is. And I think we're in a very similar situation today. My grandma probably knows the phrase chatBT. Now I can say chat GPT. That's my grandma. My grandma's got it. I know. And what a cool thing to happen for something that was historically so poorly looked upon. Like chatbot almost had like it was like people immediately associated that with a terrible experience. And now suddenly just like almost overnight, the dynamic has just been completely flipped on its end. 180 and now suddenly people are just like, they're creating accounts. And just to try out this technology and explore and utilize it, there's so much new found interest and actually just experiencing this. Now I hope that momentum keeps up. It probably won't hang around forever, but that is certainly going to shift the dynamic in terms of just overall receptiveness to utilizing this technology whenever it interfaces with a customer. 100%. And that's a really good call out that chatbots for a long time, you look down and I was like, oh, it's such a pain to have to deal with that. It's like a phone treat, right? But now the way that technology is going is actually becoming something people want to use, because it maybe is particularly reliable or consistent or can speak in their language or something. I can tell you like at the, oh, sorry. It's a good endation. Like I love it when I go to Amazon and I have to return an item. If I don't ever have to interact with a human, that's wonderful. It takes care of everything for me, just using all of Amazon's data, like takes me less than a minute. It pulls up my last few orders. That is great. And like the best AI for today is the AI that you don't know exists and just feature enhancing. Like you don't even identify that this is AI, work behind the scenes. All it's doing is just making your life better. And like that to me is like very much in the now near term, like our immediate aspiration is how do we make people's experiences better with the technology that's now available to us? It's kind of like we were talking about earlier, how there's all these different aspects of AI that can help the agent, that can identify gaps, that can translate things, can do a lot besides just be a bot that's available 24/7. The question I was going to ask is that we've touched a little bit on some of the trends that Zendesk brought out in our CX trends reports earlier this year. Jared, I'm curious, what are the things that you're really seeing as new expectations from users as AI has just started to come onto the scene and influence them? What are you hearing from customers? And what do you think is their most looking for in those improved experiences? Yeah, I think just the expectations have been reset and there is a new bar. Like for example, I could job a worker integrating AI everywhere we can within our product and now just to provide a little bit of context, in case you don't remember from where we started, we have a website building tool for example or an email composing tool. AI is now in the conversation virtually everywhere. How can we use AI to help our customers get started composing that email? How can we take some of that context that they provide us in the beginning and essentially deploy all of that work for them that they otherwise would have spent hours, days, weeks, months, sometimes even more just to get set up. Like it's part of the conversation everywhere and it realistically should be probably the biggest changes that I'm seeing is just that reset bar in terms of expectations from the customer base. But had any thoughts on that or things that you've seen from Zendos customers? Yeah, I totally agree and I think actually Jared, your example a couple minutes ago is a great demonstration of what comes to mind for me which is personalization. If you already know something about me, you know my last five orders and then I'm probably reaching out to you about it or to give another example, let's say virtual assistance aside, we want to send some kind of targeted educational content to customers who are not moving past a certain set up step at Zendosk or something. We should know what they've already tried. We should know what they have already done or maybe even ideally where they got stuck. I think the expectation that you would use what you know to make their experience better is really strong and particularly that you won't ask a question if you don't need to ask it. I think I heard Kate hint at that in last week's episode is a lesson that we learned that if we can detect information or if it's something we have that we can call up somewhere from the back end, then we should never ever be asking the user to give us that information because it's effort that they're not expecting these days to have to give fairly so we have the information we should be using it. I agree with that concept. So I think that's the biggest one. And the other thing I'll just say is that I loved what we were talking about with this sort of seismic cultural shift of Jared not only does your grandmother know what chat GPT is, she probably has a positive chatbot experience in her recent memory just like yours with Amazon. And I think that's still pretty shocking to me at times. I think we recognized in our org that as we got started with technology like this, we were fighting a little bit of an uphill battle. And so we just don't take for granted that every conversation with our bot, for example, is an opportunity to build trust or break trust and committing the time to preventing dead ends or seeking those gaps, identifying opportunities for improvement on a regular basis. I think we really want to just acknowledge that expectations for a good experience are maybe higher than ever. Maybe a couple of years ago, everyone was like, it's probably going to not be good. Now they're like, it sure should be good. So that's definitely a change in attitude that we have definitely tried to react to by just doing what we can to consistently ensure those good experiences and build trust with every interaction. I love that as our time is coming to close here, I'd love to give you each a chance to think of one thing that you would leave knowledge managers with that they should be thinking about or looking into related to AI and how they serve their customers. You can take a moment to think whenever you're ready, jump in. You don't get the jeopardy sound? Ooh, I should bust it out on my keyboard over here. Yeah, yeah, yeah, yeah, you're going to do that. All right, this thing I got to learn now. Maddie, do you want to go first? Do you want me to? You can go first. Be free. Okay. I would actually, I want to be a futurist for a moment here once again. You're before it. You have our blessing. Okay, good. I think the Holy Grail, the mecca that is hopefully coming downstream very soon is an element of support that's maybe like that. Do you remember that movie that came out in the early 2000 called Minority Report? I still want that computer where I can stand in my room and wave my arms. That would be equally cool. I don't know about the transparent screens and the hand detections is definitely possible. The Holy Grail for me is when we are solving problems before they actually become problems. When we are detecting the situations that the customer encounters and we're utilizing pattern recognition to identify when a problem is about to occur. They just, like we think about this very iteratively over the course of the last several years where we're like, oh, self service. Let's answer questions before they speak to a human. Like, the next phase of that are really the better approach to this is just let's stop the fires from occurring in the first place. And like that to me, I don't know if this is necessarily something actionable, the knowledge worker could be thinking about. But maybe just, yeah, think about, think about how do you identify the problems that yourself serve resources? Maybe they're currently in like a written medium or even a visual medium. But what would that look like to potentially solve a problem before that problem? even exists when you're not. You're no longer required to explain a situation, but you're there to potentially solve it well ahead of that problem ever occurring. That's a really fascinating idea, and where that immediately took my brain, was thinking about like, voice of the customer and product feedback and how AI is going to play into those spaces as well, and maybe is able to analyze. Here's your top thing that customers ask questions about. If you could suggest things where if you fix this in the product, you wouldn't get these questions anymore, or it could start to identify things and really help product teams even prioritize whether it's an improvement to a process or a product itself. You're making light bulbs go off everywhere, Jared. And Chloe, your answer is a great segue for mine as well, too, because I was thinking back. When I think about my career thus far in knowledge management, our strategy for years has definitely been to get to the root of what makes a customer need our help. Where do they need our help? What caused it? And what will help them resolve it? And how can we make that happen? And I think just thinking through so many of things we talked about today, clustering models that can show you where the gaps are in your knowledge base, or can show you what your customer contacts are about today, all the way up to new technologies that you could use to deploy these solutions or get answers to people in a new way and make them findable in a new way. I think really the possibilities are endless to augment what we've already built. As he said, Jared, it really is augmentation of what so many of our teams have been doing for years. And really, I couldn't agree more. I think that just as much as what our customers contact us about in tickets or what they say about us on social media, I think what they look for on their own is another voice of the customer. And I think the more we can leverage these tools that help us better understand that journey or improve that journey, we really make a difference on their overall experience with our businesses as well. So that's just really exciting to me. And I think it really makes that the potential of what we can offer in self service so much greater. Sounds like there's going to be a whole lot that we can do to improve customer experiences, to make things more efficient, to make agent and knowledge manage your jobs a little bit easier and even more exciting and interesting, not that they aren't today. But it sounds like there's just so much that's coming. I really look forward. I think that we should all get together again in six months and listen back to this episode and comment on where we are at that point and see where the world has taken us. We'll probably be surprised. We'll probably already be like halfway there. Absolutely. Well, thank you both so much for joining me today. Jared, Maddie, it's been a real pleasure to think you're both with you. Thank you for having me back. Absolutely. Well, I hope you learned something today. I sure did. What a wealth of knowledge, Maddie and Jared both shared. I hope that you will join us for our next episode as it's going to be a really special one. It was recorded live recently at the Zendesk Relate Conference in San Francisco. You'll hear from five amazing panelists that I got to speak to you, including Daniel Evans, Vice President and Chief Information Officer Honeywell Connect, William Abrams, the President of Distributed Products Division at Medline Industries, Sarah Bernardi, Chief Customer Officer at Dandelion Payments, and from Zendesk we have Teresa Nania, Senior Vice President of Customer Success and Kate Cohane, Head of Customer Advocacy. We had a lively and insightful conversation with these industry experts on what it means to be customer obsessed and why it's so important to put your customers at the center or in the front of everything that you do. We'll talk about strategy, current trends and customer experience and how AI plays into all of it. You won't want to miss this one. Until next time, I'm Nicole Saunders for Zendesk. This podcast is presented by Zendesk delivering smarter experiences across the entire customer journey for both customers and agents with industry leading solutions and expertise. Zendesk, the intelligent heart of customer experience.