Some of the most innovative brands in the world are using AI to create a service dividend - Interview with Tom Eggemeier of Zendesk

0:00So welcome to the next edition of the PunkCX podcast. With me today I have Tom Eggermeyer who is the CEO of Zendesk but also a member of the board. Prior to Zendesk, Tom was a partner of the private equity firm, Premira. And prior to that, Tom was the president of Genesis, a global leader in the Omnichannel customer experience and contact center software space, where he was there for more than 10 years. So you can see from his experience that he's been deeply involved in the service experience space for years. So Adrian, you're setting me up for failure there with giving my CV and saying I'm an expert in CX here.

0:35Well, you know, I just, I'm setting the bar high. That's fair. That's fair. And it's also kind of good to see because we spoke a few months ago when I was at the Relate event. So anyway, how are you doing? Doing really well. I've been traveling a lot right now. I'm in a hotel conference room in Los Angeles, visiting some customers yesterday and today. Last week was in Europe visiting customers. So doing well and getting the message out of what Zendesk is doing with AI right now. Fantastic. So before we dive into the kind of questions, because there's some things I wanted to ask you about some stuff that went on today, actually, at the AI Summit.

1:08But anything you want to add to the intro that I missed out? You want to set the bar even higher, Tom? I don't think so, Adrian. You did a really good job at that. The only thing is that we're going to have to do one of these sometime in Edinburgh instead of doing them on Zooms or at Relate. I will. We should do that, and I will organize the whiskey. All right. That sounds good. So let's get into it, shall we? So I know that you've just had, at the time of recording, this is October the 8th, and you've just had this virtual AI summit. And I wanted to ask you, can you give me a flavor of some of the highlights that were announced at the summit?

1:43Because there were actually, there was a lot of them. I mean, for almost like an interim event between kind of different customer events, there was like a lot that came out. And I thought, blimey, there's a new stuff launched today. And I want to understand what are the highlights from that. But how is that different from what you announced at Relate back in March? Yeah, perfect question. We announced a couple things today that I think are really important. I think you're spot on, Adrian. Um, the pace of innovation right now going on AI overall, and particularly in Zendesk is just, you know, it's really hard to keep up for our customers, for people in the CX space, even for our employees that we're spending a lot of times training because there's so much great new innovation going on.

2:24We announced a couple of things. Number one, we let everyone know by the end of this year, we're going to be at about$200 million of ARR in AI. That is not allocating part of our existing base where we drop a feature and call in that AI. That's with our new AI agents, co-pilot, and quality assurance products. We will finish the year also at about 20 ,000 customers using our AI. That's both paid and unpaid. We've got some really cool freemium products or really free products, part of our suite, things like our tentative search on our knowledge base. So that was just kind of some numbers to tell people.

2:59I guess there's a third number as well, that we're going to do 800 billion API requests this year. And we're on pace for probably one trillion next year. And I think those three statistics just tell you how quickly AI is being adopted and it's being adopted because there's a lot of value. On the specific product announcements, I'd probably put them in three or four different areas. Number one, AI and automation advancements. We launched our early access program for voice AI agents. It's a natural speech, of course, that resolves complex issues without escalation. Our knowledge connectors and knowledge builder that we talked about over late, they're now generally available.

3:38And we've had some really cool enhancements to those things like, you know, creating the knowledge articles based on your tickets and based on your industry and stuff without you doing anything except to edit them. We've had some really cool stuff. I hope we get a chance to talk about. We acquired a company called HyperArc and really cool reporting analytics and advanced insights from them. we announced, and I'll go into detail there. Copilot early access admins. We've been really focused on human agents and AI agents, particularly AI agents, but those admins are so important. And we launched admin copilot.

4:12We also went GA on no-quote tools like Action Builder and App Builder that are generally available now. So that is just like one section of what we announced. The other, the second section is new contact center capabilities. I think you might remember that we made an acquisition of a voice contact center company, and we put in video calling and screen sharing for early access there. And the voice AI agents are now also integrated into the Zendesk contact center to deliver seamless, intelligent conversations. And finally, third, we launched a bunch of stuff on employee service advancements, IT ITSS asset management early access is now natively integrated, providing full visibility into company issues, device issues for quicker IT resolution.

5:01We launched service catalog and some Microsoft copilot integration. So it's bringing directly, you know, that integration into Microsoft 365. I could keep going on, but I think this is what it's all about, Adrian. On your podcast and just the market, things are going so quickly right now in innovation. I mean, I tell you what, I was, I tuned into the summit and I was kind of, I was checking off, I was writing down this and then this and then this and this. And I was a bit like, blimey, it feels like almost what you'd expect at an annual event came out almost like it felt like an interim event.

5:37And I was like, wow, that's crazy. So, I mean, you've been hard at it. That's kind of great. And some of those numbers are truly impressive. But the one thing you didn't mention, but I wanted to kind of, to ask you about was the year you talked about in the summit, they talked about the integration of GPT-5 into your kind of offering. And I wanted to understand, you know, why you took that step and how does it strengthen your, your offering? Yes. We really think having a really close relationship with open AI and chat and being, having an early access. So, you know, when it launched, I think about six weeks ago, chat GPT-5 did, we had already been working them for months and months and months.

6:13And interesting, I think two or three days ago, I'm losing track of time, we got an award from them that we actually surpassed one trillion tokens with them. They've had less than 20 companies in their history do one trillion tokens. And again, just kind of like a moniker for how much we're working with OpenAI, how much we're integrating with them, and how much our customers are using our AI. The fact that we're passing so many tokens and using so many tokens, I should say, with OpenAI. They're also doing a great job just from a partnership perspective. We had Ashley Kramer, this chief revenue officer, OpenAI, as you saw on our AI summit.

6:51We're working with them from an architecture perspective on how do we go improve. And some of the highlights of chat GPT five performance, because we've really seen some really cool stuff. Number one, we are seeing 20 % less passes to a human agent when we're A-B testing chat GPT five versus chat GPT four O in terms of our AI agent. So much better resolution rates. Second, doing a lot better on multi-intent scenarios. And so that's, I think, one of those tough ones. When it's a single intent, ChatPTT 4.0 integrated with Zendesk did a great job, but we've really worked with them closely so that our product that's built on their large language models is now getting those multi-tent scenarios right over 65 % of the time.

7:41Third is simple tasks are going way up right now. So we're seeing success on simple tasks 95 % of the time with ChatGPT5. I could keep going with other statistics. One more maybe. The app builder tool now can update conversation prompts three to four times more often every minute. And that makes coding 25 % to 30 % faster, speeding up development. So we're excited. And this was not just like the last six weeks, as I mentioned, Adrian. This was like the last months we've worked with them. And we think chat GPT-5 is a huge advancement for our customers. Awesome. I mean, so you like, you did, so you alluded to it.

8:21I mean, that the things are going so fast right now, but, and there's almost, there's a danger in the market that technology is moving so fast and they kind of people are like waving kind of into the, in trying to go like, hello, I'd like to get on that bus. But you also kind of said that 20 ,000 of your customers are using Zendesk kind of AI. And I wanted to get a better understanding of where are they seeing the most success and where's the kind of the best adoption? Because actually that's the kind of the clue. Because I think there are, quite frankly, quite a lot of brands that are going, what the heck is going on?

8:54And how the heck do we get on that bus? Yeah, so maybe two or three points here, Adrian. One, just a statistic to demonstrate how quickly AI is moving. As you mentioned in your opening, I was on a voice contact center company for a number of years, Genesis, great company. And 2010 circa, there was about 16 million premise contact center seats out there. 2025, there's still like 5 million premise, about 10 or 11 million to move to the cloud. So by 2030 or 2035, all of those are probably going to get to the cloud, but that's going to be like a 25-year cycle to move 16 million seats from a premise model to a SaaS model.

9:37We think this is going to be like a five to seven-year transition in AI where customer support, and so it's going to go way faster than that transition. Just to put it like a stick in the ground on how fast things are moving right now. What we see are customers. I was with a New York customer in FinTech last week where they're up to now 80 % automation rates while their customer satisfaction. And it's a two-sided. It's really they've got a couple of business lines. And while their customer satisfaction is going up. And so we're seeing things like that. We've seen our customers, a lot of those 20 ,000 customers that I talked about have turned on our genitive search for knowledge base.

10:20So think about a Google, you want to go self-service, right? As a consumer, a business or employee, you type a question in. In the past, you got those 10 blue links, which were kind of not that helpful. Now you get the exact answer like a Google Gemini search. And we've seen automation go up by 10, 20, 30 % on things like that. So our customers are starting to see value on getting automation rates up different way or productive productivity up with their human agents. Awesome. I mean, because I think it's, you know, sometimes it's a, this is not a go from a standing start to a run sort of thing.

10:57This is like kind of like crawl, walk, run, just get going sort of thing. I think that's the thing. And Adrian, on that point of your mind, like, I really think that's important. And like one of the things I think we've noticed at Zendesk is we need to get our partners or our services team more involved because a lot of our customers are saying, there's so much coming at me. How do you walk me through this? I think the Zendesk value proposition was beautifully simple. You could, as a customer, you could implement on your own. You could maintain on your own. You could optimize your own. Now, you can still do that, but because AI is moving so quickly, we're investing more in partners and services so that we can help our customers go on that journey.

11:37And we kind of see three parts to that journey. Number one, you just go look at knowledge and conversations, and we see customers go very quickly to 30 % automation rates through that self-service, generative search, and what we call AI agents essentials, where you're just using knowledge. Second kind of phase is go, how do we go analyze tickets and go take that knowledge and go from like 30 to 50 or 60%. The third part where we're guiding people through the journey is how do you go start using the 800 billion API requests? So every customer has part of those requests, plugging into other systems and connecting other systems to actually make action.

12:18So it's like conversations, tickets, and then the third part of that is really plugging in. but this is a new muscle a new journey for everyone and so i think that's the opportunity but the challenge as well fantastic i mean i think that's absolutely right and it's about get started but then build up your confidence your capability as you go um but the you mentioned earlier as well about hyper arc and i heard about hyper arc on the uh on the summit and i was like oh i haven't heard about that that sounds interesting tell me more about that and so because it's apparently combines AI insights with human expertise and kind of how does, can we, how that works kind of talk?

12:55So that's a recent acquisition, right? So Rikun's acquisition and the founders of HyperArc are just really fantastic about thinking about how you're getting into data from the old kind of data reporting analytics into real insights. So I'll give you an example how we've got some really cool demos that we should share with you and your audience, but it's getting to a natural language basis. Well, so you say, hey, my automation rate over the last week went from, or this day, went on average from 80 % to 65%. You can do natural language. Why did this happen? What regions it went into? And so you're asking the AI prompt hyper arc these questions, and then it's coming back with the data and insights.

13:35This is only happening in the Asia Pacific region, or it's only happening in Scotland, say, whereas Wales, it's doing really well. And so you're able to keep drilling down, but it's not, you have to understand the spreadsheet. You have to understand in the reporting structure. It's really a natural language way, an AI way. And some of the developments that we have on the roadmap is you ask these three questions and the hyperarch might say, you might want to think about these two questions and these two insights. And so it's not going to just be the prompting, but proactive. We see you're interested in why your automated resolution rate went down.

14:11We think these are the three insights that we've drilled down in from a hyper. So that's the direction we're going. We think that's going to like really leapfrog us and our customers, most importantly, to get better insights, data, and reporting from an AI first, you know, architecture and way. Awesome. I mean, that's kind of great because, you know, it's sometimes kind of digging around in dashboards and spreadsheets and data and stuff is like a hard and kind of onerous task. You don't necessarily want to pass it off to somebody to go, tell me what that means and uncover the insights. You want it kind of like, you know, at your fingertips.

14:40So that sounds really - My team is really excited about this because every spreadsheet or dashboard I get, I ask like 37 questions. And so they're hoping that 37 questions can go down to two or three. And my team is going to be thrilled once we roll HyperArc generally available out. Excellent. And also the other thing that you can announce, which is, I guess, is pretty new as well, and is the launch of the voice AI agents. And the thing that I think about, because the activity in the voice AI space has exploded over the last year or so. And because the platform, I've got a sticker here, the resolution platform is all about resolution, right?

15:21And I just wanted to better understand what you're thinking and how you're going to ensure that they resolve issues accurately and empathetically and don't just become this other, if you like, roadblock, which then just kind of allows you just forces you to escalate. Yes, so a couple things. One, I think the technology has really, really changed to the positive. I'm one that hates getting in an IVR, an interactive voice response system, and I very quickly get frustrated and hit zero because it's a decision tree, right? It's like, if yes here, if no, and you just keep going through that decision tree.

15:57My experience with testing our voice AI agent, it does feel like you're talking to a person. And so I think the technology is making a huge leap. It's not perfect yet on voice because latency matters. Whereas when you're doing with data, it's really easy, asynchronous or near synchronous. And so I would tell you, it's just a massive, massive jump over like IVRs and other voice bots in the past. And so like that's one, if you played around with it or your listeners played around with it, Adrian, I think they would really, really be excited on tone, empathy, the low latency, the accuracy of answering questions.

16:40And it feels like a human conversation saying that and why it's important for us is we believe our resolution platform has to have every single change. So our customers are telling us they want one company to give them everything from generative search to AI agents to work workforce engagement management to voice capabilities to you name it, core ticketing case management. Right. What we recommend our customers, this is part of where Zendesk is evolving from a, what do you want to do? What do you want to do customer to we recommend is we think you need to have that operator option with a voice AI agent so that you just don't keep looping around just like people did in IVRs.

17:22And if you get frustrated with something, we think the technology is way better. You can go say, I want to talk to an operator. I want to talk to someone live. And And so that's like one of those things that we recommend to our customers, whether they implement our voice AI agents or someone else's. And I think what's really interesting as well is because it's not just kind of the voice AI agents on the front end. You're also using the voice technology on the kind of on the agent side to do that kind of proactive real-time kind of guidance. So it's listening to the conversation, picking up intents, picking up sentiment, and then drawing from knowledge and then providing kind of like suggestions in real time as well.

17:54So it's like a two-sided sort of solution. Totally. It's a two-sided solution. And we're even doing things like we're going to be ultimately, it's not yet, but like based upon sentiment and things, you're going to route it maybe to a different human agent. Someone that's, you know, really good with someone that's upset about something. You know that with, you know, predictive routing and omni-channel routing or predictive routing, who you should go within your available agents to route it to, who's best dealing with, you know, upset customers or upset businesses or upset employees. and exactly you're really getting insights real time as a human agent which we think you know improves empathy improves accuracy and makes them more productive because they're getting suggested answers to them versus them have to type in and reactively ask questions to the system awesome so i know you're a busy man and i'm sure you've got other things to do so let's kind of like let's go rattle through the next one so any kind of like examples of brands that you'd like to give a shout out to in terms of these guys are doing some great things, you know, based on kind of some of the new capabilities that you're announcing.

19:00And this is just because I was in New York a couple of weeks ago. There's another two-sided marketplace in New York. So I talked about a fintech company, but I'll take a two-sided marketplace where what they're doing is they're taking their automation rates up pretty substantially and have a really great plan to go through that conversations, knowledge, ticketing, interaction journey. And they are absolutely reinvesting the savings they have onto how they get more sticky in a two-sided marketplace with their sellers and buyers. And so I think one of the things out there is this is going to be a cost savings exercise.

19:35Some companies are going to take the cost savings to the bottom line. I've actually seen some of the most innovative brands in the world, and this brand sticks in my mind, they're saying we're going to go with a service dividend and we're going to make the service experience even better because we think it's going to create a loyalty loop and create our, you know, this is again, a two-sided marketplace, our buyers and sellers even closer to our brand. And this is a CEO of, you know, multi-billion dollar enterprise in terms of revenue. And that's how people are thinking where I think there's more about cost six months ago.

20:08It's more about how can I go take that automation to increase customer satisfaction? How can I take that automation to go work on things that we've always wanted to work on to increase service levels? I think it's a great point. I love that idea of the service dividend because I've always said that when I've been thinking about this is that the potential and the possibilities of new technology, it provides choice. And it's choice that you can choose to save money or to reduce the number of resources. but you can also make another choice, which is you can take the dividend and pocket it, or you can take it and you can use it somewhere else.

20:44Exactly. And so that's a great example. I love that. So, Tom, anything else you want to add before I ask you some quickfire questions? All good, Adrian. Okay, fine. Quickfire questions. Here we go. So I always like to try and boil things down for people. So here's my question. What would be your best advice to someone listening and who wants to improve their customer experience? I would say, make sure you work with your vendor really, really closely on a joint roadmap and really put your North star and envision what you're trying to accomplish up front and march towards that, you know, with a joint engagement with your vendor.

21:23Perfect. Or your partner. I hope we were partners. Yeah, absolutely. And so now for a punk one, because it's a punk CX podcast, That's obviously. Which company or brand do you think takes a more punk approach to customer experience and why? Yeah, of course I'm going to say Zendesk. Shameless blog. I know, but I won't continue with the punk's shameless approach. It's kind of interesting. I was in Germany last week with some of our customers. And some of our customers are doing just some really innovative things. a global brand from a gambling perspective, again, taking that punk attitude of we want to make sure that our customer inquiries are decided in a couple seconds and they're complex inquiries.

22:08And we want to go make sure that we're getting just really top CSAT scores. And so, again, some of these brands, they think they're getting a little bit of a competitive advantage by working with Zendesk and automating and having service dividends, which is really cool. but let's just say a global or particularly European-based gambling brand is really thinking about it in a great way, how they give that service dividend again back. And my final question before we wrap up, and I always like to, we're in a, the world's in a funny place right now. There's all sorts of doom scrolling going on. I always want to end these conversations on a good news story.

22:40So Tom, can you tell me the most interesting, positive, or exciting thing that you've seen in the last week? Well, the thing that's made you smile. Last week, I was in Germany at Wiesn or Oktoberfest with 50 of our customers and a bunch of Zendeskians, our fellow employees. And just like it was great seeing these customer innovation stories. It was great seeing my fellow Zendeskians really excited about AI. And quite frankly, it was great sharing a pint or not a pint, a liter of beer with our customers. Exactly. A nice big Stein, a Krug, a Moss with our customers. That made me smile. We have an event we do a lot of times with customers at Oktoberfest.

23:23And I think it's just a great way where they see four or five hours of our technology. But then we have a good time talking about the technology of our beer. Awesome. What a great story. And I can imagine possibly quite a lot of tipsy people as well. But so, Tom, that's all I have for you today. I just want to say thank you for sharing your time and your insights and your expertise with me today. That's been great. Thanks for having me again, Adrian. I look forward to one of the times in the next year or two doing it in Edinburgh. Perfect. Thank you.

23:52Wow, what a great interview. I hope you enjoyed it. I know I did. Find out more about me and the work that I do at adrianswinsko.com. Do leave a review on your favorite podcast platform. And if you have any comments, feedback, or questions about the podcast, then feel free to send me a message to podcast at adrianswinsko.com. and do tune in again. Thanks very much.