You probably feel like you know your customers pretty well, but you can’t know everything, and there always seems to be a new question that you or your team are unprepared to answer or respond to. How do you identify those gaps before you get a tough question from your customer?
Today we’re going to talk about the transformative power of knowledgebases in customer support with Sophie Wyne, CEO of Ariglad. We’ll discuss how a well-maintained knowledgebase can benefit both customers and internal teams and how AI plays a crucial role in optimizing this resource.
Sophie Wyne is the Co-Founder and CEO at Ariglad. After participating in the prestigious Y Combinator program in San Francisco, Sophie raised $2M of capital in Ariglad’s Seed round. Since launching the first AI knowledge base tool of its kind, Sophie is focusing on educating the CX space on the unique benefits of AI and is a leading voice in the growing AI Customer Success space. Ariglad is headquartered in San Francisco and the team is spread across North America and Europe.
Resources
Wix Studio is the ultimate web platform for creative, fast-paced teams at agencies and enterprises—with smart design tools, flexible dev capabilities, full-stack business solutions, multi-site management, advanced AI and fully managed infrastructure. https://www.wix.com/studio
Attend the Mid-Atlantic MarCom Summit, the region’s largest marketing communications conference. Register with the code “Agile” and get 15% off.
Register now for HumanX 2025. This AI-focused event which brings some of the most forward-thinking minds in technology together. Register now with the code “HX25p_tab” for $250 off the regular price.
Connect with Greg on LinkedIn: https://www.linkedin.com/in/gregkihlstrom
Don’t miss a thing: get the latest episodes, sign up for our newsletter and more: https://www.theagilebrand.show
Check out The Agile Brand Guide website with articles, insights, and Martechipedia, the wiki for marketing technology: https://www.agilebrandguide.com
The Agile Brand podcast is brought to you by TEKsystems. Learn more here: https://www.teksystems.com/versionnextnow
The Agile Brand is produced by Missing Link—a Latina-owned strategy-driven, creatively fueled production co-op. From ideation to creation, they craft human connections through intelligent, engaging and informative content. https://www.missinglink.company
Transcript
Note: This was AI-generated and only lightly edited.
Greg Kihlstrom:
You probably feel like you know your customers pretty well, but you can’t know everything. And there’s always seems to be a new question that you or your team are unprepared to answer or respond to. So how do you identify those gaps before you get a tough question from your customer? Today, we’re going to talk about the transformative power of knowledge bases in customer support with Sophie Wyne, CEO of AriGlad. We’ll discuss how a well-maintained knowledge base can benefit both customers and internal teams and how AI plays a crucial role in optimizing this resource. Sophie, welcome to the show.
Sophie Wyne: So excited to be here. Thanks for having me.
Greg Kihlstrom: Yeah, looking forward to this. Why don’t we get started with you giving a little background on your journey to becoming CEO at AriGlad.
Sophie Wyne: Yeah, so I actually worked in customer success prior to this and worked on the company knowledge base. And before that was really just working customer service since I was about 14. And so I know that the customers often want to find their own answers. And my co-founder is very much technical in the AI space. So we combined those two backgrounds to create what is really the first AI tool of its kind to make sure that If customers want to self-serve and look for answers in the knowledge base, that answers will be there and they can self-serve successfully.
Greg Kihlstrom: That’s great. Yeah. And you touched a little bit on it, but I wonder just before we dive in further into the some of the questions, could you touch a little bit more on, you know, what Ariglad does, like who’s who are your customers and things like that?
Sophie Wyne: Yeah, so we basically integrate with data sources. So we integrate with your ticketing system, your Slack, release notes. Basically, the AI learns how your product works and the types of questions that you get from customers. And then it integrates with your knowledge base and it identifies gaps. So it creates new articles if there should be one. if there’s enough questions from customers to warrant it, and updates existing articles, merges duplicates, basically just cleans up your knowledge base and builds it out where it needs to be.
Greg Kihlstrom: Great, great. So let’s dive in here. And I want to start by talking about the importance of a comprehensive and robust knowledge base and its impact on customer satisfaction. So could you speak to that a little bit? Why is the knowledge base so essential for providing excellent customer support?
Sophie Wyne: So there’s kind of the, I guess, qualitative and quantitative answer. So I would say from both of our experiences, I’m sure we’ve both been on the consumer side where we have a product and it broke or it’s not working or what have you. And then you’re trying to find that answer online. And maybe you find what should be the right article, but it has the incorrect information or you just can’t find an answer. And so you kind of have to go through the rigmarole of reaching out. I’ve definitely experienced that at least. And then there’s more of the quantitative side, which is that there’s been quite a lot of research into this space that has produced a lot of data that supports just how critical knowledge bases are. According to a Forrester study, 81% of customers look for their own answers first. So they’re actually seeking to self-serve before they reach out to support. And there’s just a lot of research, also supporting just how hard it is to be in customer success and support today. Customers have really high expectations. A Zendesk study came out recently that said around 50% of customers feel stressed and exhausted after interacting with support, which is not fair on support. They do an amazing job, but People are just stressed out. 53% of Americans experience high levels of stress on a daily basis. So it’s often just the last kind of straw for a lot of people. And so for us, we really just want to make sure that, again, we’re reducing just how many customers are reaching out in the first place. Customers are happy. Customer success and service has more bandwidth. So it’s really a win-win.
Greg Kihlstrom: Yeah, I mean, it makes sense, because there might be other reasons you reach out to a company. But when you’re reaching out to customer support, I mean, generally speaking, there’s an issue, if not a real problem, and sometimes a very pressing problem. So, you know, I know, personally, and I’ve said this plenty of times on the show before, I’ll do almost anything to not pick up the phone and call someone. So I’m one of those people that self service, just I’ll do anything to find it myself, and everything like that. And I literally yesterday, not even prepping for this show, but literally yesterday, access to knowledge base. And you know what, it was out of date. And it wasn’t anything near what it was, like what I was expecting to see and stuff like that. And so, you know, the frustration is real. And I, I didn’t even pick up the phone and call. I just kept hunting to see, okay, well, maybe somebody said something on Reddit or some somewhere else. So, you know, that’s how, how does your how does your platform help organizations maintain, you know, effective knowledge bases, maybe, maybe to solve some of the problems that I just described in my own, in my own experience, but you know, just not only maintaining, but also, you know, developing in the first place.
Sophie Wyne: For sure. Yeah, I mean, it’s so hard to keep knowledge bases updated. You know, I really do have to emphasize that, especially for, I mean, tech companies, but even non-tech companies. But if we take tech companies as an example, you know, we’re working with some customers that are releasing new product releases twice a week. And so the product is just moving so fast. And so you really, if you don’t have a full time team that’s working on maintaining the knowledge base and, you know, replacing screenshots and making sure that every piece of text is updated, that every piece of the tool is reflected in the knowledge base, it is just going to become out of date so, so quickly. And it can actually create just this habit spiral where if a customer never finds their answer in the knowledge base and they can never find it, they’re just going to create a support ticket. And that’s going to cost more and more money. It costs, I think, over $13 for a B2B company for every live service interaction. So that just adds up super, super fast. But yeah, on the technical side of how we really do that. So in this example that you just mentioned, where there was a, you know, an article, there was probably the title was right. And then the content was incorrect. So the way that article out of works is it basically looks at, okay, first of all, has any have any customers already reached out about this? Have they said, Hey, I have this question. And this, you know, the support agent is able to give an answer. And if the question correlates with the topic of that article, the AI is going to see, OK, are the answers that the support’s providing not correlating correctly with what’s in the knowledge base? And if not, it’s going to basically make those edits in the knowledge base and say, OK, there’s 10 support tickets. They’ve all basically provided an answer that is not what this article is providing. And so we would suggest these edits. And then support can basically click Approve and say, yes, that looks good. And the edits will be made.
Greg Kihlstrom: Nice, nice. So then, you know, there’s a cost to all of this, you know, as far as there’s time and effort and, you know, even platform costs and things like that. But when organizations are looking at success and, you know, some of that is ROI, but also some of it is customer satisfaction, right? So, you know, how do you recommend that? your customers, you know, measure their investment sort of from a customer satisfaction standpoint.
Sophie Wyne: Yeah, I mean, again, there’s so many different angles to look at. So a lot of our customers will be focusing on, you know, NPS scores and making sure that those are staying high. So if you are focusing, for example, on tickets being handled really well every time, you’re going to need potentially some support for people doing support. So agents with AI tools or even agents looking through the knowledge base. So for them to provide accurate responses, they need the resources that are going to provide accurate responses to customers. Then we have the customers that are not going to want to wait at all. They’re going to want an answer as soon as possible. A lot of data has been coming out of how expectations, especially with AI chatbots on the rise, the expectations for super fast responses for support has just been increasing. And if you don’t have a robust self-serve operation, all of your customers that have questions are going to be going to support, and it’s just going to completely block you up. And you’re going to be hiring more and more people to try and put a Band-Aid on the problem, and potentially invest in more AI chatbots, when really A huge amount, if not the vast majority, of those customers in your support channels looked for those answers first. They tried to self-serve and would have been happier to begin with. I mean, I’m sure if we take your example, yes, even if you went to support after not finding answers and you had the most amazing experience, your experience with that product would have been tainted somewhat because you weren’t able to find the answer in the way and format that you wanted in the first place. So it’s kind of a, you know, It’s not a great experience for both sides because really support can’t do anything about that prior experience you had where you couldn’t find that answer. And, you know, support also would have benefited if you had just found that answer and then left them to have a bit more bandwidth.
Greg Kihlstrom: So let’s talk a little bit more about those those internal teams, because, you know, as as you rightly mentioned, you know, there’s there’s a lot of great customer support teams and agents working really hard. And, you know, I’m as much as I personally don’t like picking up the phone and talking to people, you know, nothing, nothing against them. That’s my that’s my stuff. We’ll just call it that. I’d prefer to self serve and stuff. But that said, you know, I’m an optimist when it comes to I think most people want to feel fulfillment in their work and want to do a good job. You know, I’m sure there’s some people out there that don’t feel that way. But I believe the vast majority of people want to do a good job. And yet, you know, these teams are, they’re, they’re stretched thin often, they’re, you know, there’s too much information, there’s not thorough information and stuff like that. So, you know, looking at the from the internal team perspective, you know, how does a comprehensive knowledge base, help those teams and help some of those things that I that I touched on there?
Sophie Wyne: So we first of all want to make sure that the customers that want to self-serve were successful and are not clogging up your support channels. But if we put that aside, from an agent perspective, this is kind of the hot topic in a lot of customer success and support leadership verticals. where they want to use AI and AI kind of chatbots or AI response assistants to support these agents to provide the best experience and as quickly as possible. So, you know, if you, for example, let’s take Zendesk, there’s a, an app that you can now have in your support channel where when you get a question in, you kind of have like a co-pilot where the AI will provide a, what they think should be a great response to this customer question. And you can basically take it, you can make any tweaks that you want, and then you can send it off. And so it’s doing a huge amount of work, not only from just typing that out, but just finding all of those resources, find those bits of information, sourcing it for you and then presenting it to you. But what happens is these AI chat bots as well, separate from support individuals, but AI chat bots and co-pilots, they actually most of the time need to be connected to your knowledge base. And so what they really need at the end of the day is an accurate and reliable data source that is up to date with how your product works. And if you don’t have that, you’re always going to be stuck with either shoddy results from these expensive AI tools that do have a ton of potential, but need that kind of basic knowledge and data sources. and just playing catch up constantly in the knowledge base and just basically realizing a little bit too late that there’s incorrect information because these AI response systems are not providing the right information. So it’s kind of a double, I guess, a whammy of, you know, if you don’t have an accurate knowledge base, you are having more people reaching out in the first place, clogging up your support channels. And then even if you are kind of investing in your agents making sure that they have access to resources and they have access to AI response assistance or what have you. Again, if you don’t have an accurate knowledge base, it’s going to be really hard for them to do their job to the best of their ability.
Greg Kihlstrom: Yeah, yeah. And I think there’s a few things in there that you touched on that you could use as measurements. But similar to the question that I asked from the customer perspective, it’s like, how should an organization measure this? Is it all looking at customer satisfaction? Are there internal measurements that you can use to see things as well?
Sophie Wyne: So there’s been a couple of different ways. I would say customer satisfaction. I would also say there is the ability to see how popular a article is. So if an article was really popular and then it kind of went down in value, And now the AI has come in, Ari Glad has updated your knowledge base, and you’re starting to see the traction build and build over time within your knowledge base. That’s a huge kind of direction of success. And then, of course, reducing the amount of support tickets in general. depending on what knowledge base you use, we can actually play around a little bit with the tagging and making sure that we can see that correlation of customers are looking for answers to this question and they found that answer in this article. And so yeah, there’s a couple of different ways that you can see that success. And then of course, if you’re really just looking at agent success and reducing the amount of time you’re spending on those support tickets, If you’re looking at investing in an expensive AI co-pilot, which again, actually work really, really well, you are not going to see the results that they’re promising and the full potential of those unless you have an updated knowledge base. So that’s also a huge parameter for success. So taken all together, it’s really just kind of supercharging your entire support to vertical, but individually it’s showing, I would say at the beginning, it’s really just showing that customers are able to find their answers in the knowledge base and those articles are being really heavily used.
Greg Kihlstrom: Yeah. Yeah. So even, even things like, like time to solution or is that, you know, kind of, yeah, yeah. Got it. Okay. You touched a little bit on AI in a few different ways, but I wanted to ask just specifically how ARIGLAD is utilizing AI and if you could talk a little bit about that.
Sophie Wyne: So from a technical perspective, and this is going to be the dumbed down version because I’m actually not technical. This is how it’s been, you know, I’ve learned kind of through the ringer of being a non-technical person in an AI world. Basically, let’s imagine we bring on a new customer. We integrate with their Zendesk, their Notion, Slack channel. It’s basically going to be looking for, in the Slack channel, questions and answers. And the same thing in the Zendesk. It’s going to be looking for, okay, where is a question and where is an answer? And then once it understands that this is a question, this is the answer, where can we now correlate that data? Has this information been provided consistently? And so it’s going to be looking at that either across channels or maybe there’s just a ton of those answers provided over and over in just Zendesk. Either way, it’s going to be looking to make sure that this is not just a one-off answer, that this is actually really can be seen as the source of truth for this data that we’re looking to embed in the knowledge base. So once it’s validated that, it’s going to take those answers and it’s going to look at your knowledge base and it’s going to say, okay, we have all of these answers. These are all answers to questions that have come up enough where there definitely should be data for these in the knowledge base. And so it’s going to find those gaps, it’s going to create new articles, it’s going to update existing articles, and it’s going to do that on a, you know, either weekly basis or a near daily basis.
Greg Kihlstrom: Got it, got it. So we talked a few different ways about measurements and seeing improvements. What improvements have you seen that might be closely tied to using the AI for analysis and things like that? What could customers expect from that?
Sophie Wyne: So we’ve seen a couple of different ones. We’re actually working on some case studies to basically show, like you said, what is that one consistent piece of data that you can look to to see that this is working. But something interesting that we’ve seen is some of the customers that have brought us on had very high expectations for ROI themselves. And we didn’t even realize this. And it would be a few months later after using us, and they would come back and say, look, we were really suspicious, and we had really high expectations for, these are the parameters that Areglad needed to help with and meet if we were going to continue. And we did meet it or exceed them. And so one of them that comes to mind is an organization that basically wanted, they had an individual that was basically in charge of maintaining the knowledge base. Not only the knowledge base, but maintaining the back end of the kind of call center and support channel and all of the data that the support individuals were getting as they were providing answers to customers. So she had just a huge job of making sure that everything was updated at all times, and they wanted to reduce the amount of time she was spending on this in half. So they wanted a 50% reduction in time spent on this, and Ari Glad was able to exceed that. And so basically, all she needed to do was log into Ari Glad, look at all the suggestions. As you’re looking at a new proposed article, you can make any tweaks if you would like, but you’re basically approving it, you’re sending it out. And so that just reduced the amount of time she was spending on this by a crazy amount. And they were left with a really, really updated knowledge base. And that organization ended up wanting to roll this out for multiple different departments because it was so successful. that wasn’t potentially, that wasn’t exactly the data metric that we were thinking would be the most important, but now we’ve added it to our roster of different ways you can see how this is working for you.
Greg Kihlstrom: Yeah, yeah. And so, you know, I work with a lot of companies on various, you know, AI implementations, and there can be some challenges, there’s lots of benefits, of course, potential benefits, many realized benefits as well, but what are some of the challenges that businesses may face when, you know, AI sounds great to solve every single problem, but, you know, often there’s some challenges in integrating it and integrating systems and all those kinds of things. So, you know, what are some of the common challenges that you see and, you know, what are some ways to overcome them?
Sophie Wyne: So I would say on our side, one of the challenges that comes to mind is just not having enough data or not giving access to enough data. The AI, especially for our AI, has very high standards as to what data makes the cut to be integrated into the knowledge base and be proposed as a change. If we are limited as to how many support tickets we get access to, or basically if there’s just not enough data, the AI is going to be crippled in its ability to do that to the fullest extent. But we now know what that threshold is. So usually what happens is we can already tell that there’s not going to be enough data. So we basically just let the customer know, and we work together to make sure that we do get enough access. On the other side, I would say, honestly, and this is not to toot our own horn necessarily, but I went to a conference recently for a large ticketing organization, and they were really focusing on the AI benefits, the AI bots or co-pilots or what have you. And one of the interesting challenges I could just see, as they would talk about how much time it’s going to save and just how helpful this would be to reducing time spent on tickets or success rates, but the knowledge base still needs to be updated. And I would say, yes, okay, we focus on knowledge bases. I’m a little biased, but I would say this really just comes down to data sources, which even speaks to if our data sources aren’t big enough, or if a co-pilot’s data source isn’t organized into one space, that is going to be just a kick in the pants to the AI, to be frank. It’s not going to be successful. You’re not going to be able to see it work at its full capacity. And so I would say having a data source prepared and ready for the AI before you implement is just super important because otherwise you’re going to be paying for this AI tool while you’re getting the data source together and the AI is not going to be able to work properly.
Greg Kihlstrom: Yeah, yeah, I totally agree. I’ve, I’ve seen that time and time again. It’s, it’s, that’s one of the things that I, I recommend to everyone. It feels like it’s, it’s kind of delaying things because you’ve got to get the data house in order, so to speak. But yeah, to your point, it, when it works, you know, when, when you plug it in, so to speak, it works and it works a lot more quickly. So it’s, it’s worth that, that set up. Absolutely. Well, you know, to kind of close things out here, and you know, it’s been great talking with you about this, I wanted to get your thoughts on future trends here and the future of AI and customer experience and things. So, you know, what trends are you seeing that are going to influence AI’s role in some of these things that we’ve been talking about, the customer support and customer experience?
Sophie Wyne: Yeah, I would say that the trends have been pretty clear that organizations want to implement AI to take care of the more basic questions for customers. I would say a trend that I’m starting to see a little bit more is that organizations want to elevate AI’s ability to handle a little bit more complex questions and also come across as a little bit more human. A lot of organizations are even asking us how they can change the tone or how the AI is coming across and how they can really kind of mold the AI to be in a sense that like a customer support agent on their team with the same type of vernacular and way of coming across. So I found that quite interesting, but I do think that there is then on the other side organizations that are waiting for the AI to get better because they don’t want it to be associated with their team specifically. They want there to be a very direct line of this is AI and We are not completely behind them quite yet, but they will get there. So I think, yeah, I’m definitely seeing in the future that as AI continues to get better and those data sources continue to grow and they can really shine. Because I think AI in a lot of organizations where it’s, if you implement a chatbot and it was disappointing, a lot of the time it’s just not great data that it had access to. As those gaps close and AI just really is able to shine and grow to its full potential, you’re going to see organizations become more comfortable with the AI really taking on more of a, I guess, human element. And yeah, I’m looking forward to seeing how that goes.
