It cool. I think we're live. Mr. Mickey Powell. Welcome to the podcast. Thank you, Alex. It's real nice to have you. For those of you who've been living under a rock in the CS community and haven't paid any attention to LinkedIn or any of the number of groups that are out there, mickey, you've been kind of all over the place and I would say at the forefront of kind of AI in CS, which is cool because it's being talked about a whole lot. You're kind of like the AI guy for CS at the moment, which is cool. And it's a pleasure having you on here. I'm sure we'll dig into some of the work that you're doing with Update AI as a co founder, right? Yeah, co founder. And I've sensed from you, just from some of the content that you've put out is that you're not only an advocate for Aincs, but you're an advocate for using it obviously in productive ways, but then also not just using it for doing more, but using it for doing it better. So they're very accurate description, for sure. What else do the people need to know about you? How did you get to this place where you've carved out a bit of a niche for yourself? Yeah, so I didn't have the opportunity to go to college, so I moved from Central California for the folks that know where Viscelia is, it's in between Fresno and Bakersfield to San Francisco in 2009. And I was 19, so I didn't fully grasp just how bad things were because that was like my first real global catastrophe. Now I'm a seasoned pro. I've been through a bunch of them. Right. So I moved there. I started working retail at at and T down by the ballpark, which was awesome because I'm a Giants fan. So, like, to be literally across the street from Giant Stadium, which was really, really fun. And I met a bunch of people from tech companies down there, like early days of Square Dropbox, Ubisoft Yammer eventbrite, and probably a bunch more that I didn't know that turned into big companies or most likely fizzled away. And because I wasn't able to go to college, I kind of made a pact with myself as like, I am going to be a lifelong learner no matter what. That is how I'm going to get where I want to go in life. Because like everybody, I have hopes and dreams and desires and altruistic goals and hopes and less altruistic goals and hopes and at and T was a great landing spot for me needing a job, needing a job that pays fairly well for a 19 year old. But it was not the end goal. I just knew I wasn't going to be able to do what I wanted to do. And I saw this whole tech world and I was like, I need to get into that. You saw a glimpse of it and you jumped in. Yeah. So I think that that's a good indication of fast forwarding to Chat GPT, which is I'm really good at learning and integrating things really fast and I love Sci-Fi and I love futuristic things and I have a really easy time, kind of like putting aside the state of what is and thinking about what could be. And there's been lots of technological innovations in my lifetime, but I think Chat GPT was kind of just the right it was just the right time for a lot of reasons for me. Number one, it's hugely applicable everywhere, but I knew that it would be hugely applicable in customer success. I know of all the industries and all the things that I have done, customer success is what I know the most about having spent the last ten years of my career in customer success, from being a CSM to a leader to CSOPs. And the third point is, I now have a daughter who's four going on five. And when I kind of fully grasped how big this was, I said, all right, I am going 100% all in as hard as I can go personally to understand this and integrate this and leverage this in whatever way. I can to not only help myself, but to help my daughter and my family and others. Because I do believe in trying to give a lot of this away and helping other folks. Hence why all the content and stuff I put out, I'm not charging for it. Sure. And so it was kind of the right moment at the right time. And it's not lost on me that an innovation like this doesn't come around every other year. I like to say. Yeah, like, don't go chasing waterfalls like that, I guess. I don't know. Is that the right in the right term? I don't know. It's interesting. You and I were talking briefly about kids before we started recording, and it dawned on me not long ago that the skills that our children are going to have to develop as part of this whole transition is not necessarily yes, writing skills and all those things are always going to be relevant, I think, but also the prompting. I think prompting for generative AI is probably going to become a skill that we should start teaching. I think. So in the short term. I think in the long term, it actually won't be as important. There's a million new AI tools out there and I just had somebody ask me today, how do you keep up and which is the good ones? And my answer is to say this understand the underlying technology. That's the first thing you need to do. Because just like how you understand how an internal combustion engine works, you don't understand all of it, but you understand that pressing the gas puts gas into the engine, which provides power, which drives the car forward. You need to understand, just like at a foundational level how these technologies work. All the innovation that's happening right now and this includes update AI, all they're doing is taking the base technology and then layering on top their own UI and their own specific set of prompts to achieve some sort of goal for their customer. And that's still amazing. It's still an amazing innovation. But once you realize that, it kind of demystifies the million new tools that are being launched every other day. Yeah, that makes a lot of sense. I want to ask you one more quick question about your background a little bit because we actually share somewhat of a common trend because you do have some education, training background as having been a trainer in the past. And I feel like customer success people and training people all share a very similar thread, which is to say kind of that enablement, that helper, that whatnot, that DNA that we kind of share. And I was curious what part of your personality and psyche kind of drew you to, I guess both of those modalities of professionalism. Yeah, I think I've always liked to help people. I've always liked to take what I know and help impart that to folks. And it's funny, right, because strengths can also be weaknesses. The strength side of that is kind of what you're seeing from me right now with Chat GPT, taking this big scary, complex thing and trying to distill it down for folks. The negative side of that is like, sometimes I come across as a know it all. So I've been very conscious of trying to balance that. And I wrote an ebook that I just launched last week and in that ebook I tried to be very clear. Yes, right now I could be considered an expert because I know more than most people. But there's real experts out there. Here's who they are. Go listen to them. I'm going to give you my opinion and I'm going to try to make it clear when it's my opinion versus fact and what evidence I have and try to do it very humbly. The word curator comes to mind. Yeah, I'd say one of the areas I'm strong and why I got into training is I'm very good at taking all of these things and putting them into a package that kind of reasonably makes sense and then give that to people that either don't want to take that time or can't take that time or some other limitation. The second thing that really drew me to training and customers and enablement and all those things is because people need that help. And it was modeled for me. My parents are very much both like loving, caring, helping people. And so I have seen that I'm the youngest of four boys. I don't know if that matters. Sure, the dynamic of where you fall in the pecking order, but I will say this I spent a year being like a training Manager, which was kind of a combination of CS ops and instructional design and training facilitation. And it was really hard. Instructional design, especially is a very, very difficult field. It is. And I realize I'm better at more of the operational strategic work, so I kind of abandoned that path. But I'm also really excited because I think that that industry is going to get a lot of help from these new tools. Yeah, it's difficult. Instructional design is difficult. And so misunderstood, too. This is the digital customer success podcast. And I would be remiss if I didn't ask you for your elevator pitch of digital customer success. And really, the plan is this is something that I want to consciously ask everyone who joins, because I'd like to build kind of like this holistic wordmap, if you will, of everyone's take on Digital CS, whether you take that to mean scaled CS or digital CS or whatever. So what would your elevator pitch for Digital CS be? Yeah, my elevator pitch for Digital CS is it's an organizational alignment of strategy, product processes, people, systems and data to amplify and accelerate the adoption and the value that customers receive from your product. I like it. You hit all the keywords, I think. Yeah, that's right. I want to make sure that your SEO game is strong after this. SEO on point. Got it. Thanks for that. Yeah, that's good. Yeah, I like it. It's interesting because I think you were at Pulse last week, correct? Yeah. So this is a week after Pulse 23, and there was a lot of talk about not just how AI and Digital CS are helping customers, but it's also this notion of helping. Digital CS isn't just about helping customers. And enabling customers, et cetera. But it's also about supporting the CSM themselves and what it is they do and making them more efficient as well. Yeah, I'll use an example, and for those that are curious, and I think you're going to ask this question later, I tend to like to read or learn things that are more foundational and then figure out how to apply them in a context as opposed to reading very specific things and then figuring out the underlying themes. So there's a book good Strategy, bad Strategy by Richard Rumel. He's a professor at the UCLA School of Business, and I've actually read it twice, and one of the examples he uses is Ikea. In terms of their strategy. Their strategy used to be vertically integrated. They own every single step of the value chain from sourcing of the wood. Actually, I don't know, but they probably grow their own wood is my guess. Just to cut out the middleman sure. Through getting it into the house of the customer. And they ruthlessly optimize everything in that chain, from the packaging to where the screws are placed, to everything. And they have completely dominated that industry. There is nobody that is even close to ikea, right? So when I think about Digital CS and I think about Ikea as, like, the analog, we have to continue to ruthlessly optimize how customers move through their journey with us. And Digital CS is a huge part of that. Because if you're saying I can't do Digital CS, you're basically saying there's multiple methods of communication that I'm just not going to use, that my customers want to use. And it's just like it's folly. Right. We'll use Chat GBT as the example. There's no human guiding us through how to use that platform. And it was the fastest growing consumer product ever. So if you get the design right and you're solving problems for them and creating value, you will see so much like the amplification that I mentioned earlier, you'll see such a better return. The problem, of course, is that the tools that exist to do that, with a few very minor exceptions, they generally haven't been in the hands of customer success teams. And then it's like, can you get a cross functional project going to do that? And then that's a whole can of worms. Yeah, absolutely. And that alignment around tool sets that are really pointed at one specific function, like, for instance, there's a ton of sales tools out there, and a lot of them would have broad applications on CS. But sometimes it's just that transition just doesn't make it over, which is interesting. Speaking of kind of tools and really driving value within the customer, within your customer, I'm curious as to what you're currently measuring from a metrics perspective, your customers on specific to your digital programs. Yeah, so we're very underdeveloped in this area because we're a very young company. I think everybody kind of is, but yeah. Well, yes, I will say yes. Most companies I've seen are so number one, we're still trying to measure the broad things because we actually have not, as of today, started charging our customers. We're still just, like, growing, letting people use it for free, letting them know, hey, in the very near future we're going to start charging because we can't do free forever. So we're measuring the common things, right, which is like, how quickly and do they get through our onboarding? Can we optimize that? Do they start to use our product? There's a particular moment of value where when they start using our product, they're going to see the very first time they use it, how valuable it is. Do they get to that point, and then do they refer other folks? We have not really started measuring anything deeper than that. We do a little bit of event driven email behind all of that. But coming from my background, I definitely cringe when I think about everything we could be doing. And it's really just a matter of, like, time and prioritization and resources. At the end of the day, especially, I think if you're in the beginning stages of things, you really want to drive that time to value. And I think product metrics is a great way to get there quickly. And it's not realistic that you have conversations with every single one of your customers all the time, right? No, we couldn't. We're fortunate that we have over 1000 CSMS using our product. I wish I could talk to all of them all the time, but yeah, that's just not realistic. Not realistic. And what we want to continue to do is to look at other proxies to understand the value that they're getting out of the product. Yeah, that makes sense. I'd like to shift back to kind of Pulse real quick because again, we're a week out from Pulse 23 and I'm curious, obviously GAINSITE made some cool announcements from an AI front and there's a lot of other platforms that are kind of internalizing AI as a product feature or generative AI as a product feature. And I'm curious what your hot takes were in general from the two day event. Yeah, I think this is probably a good time to say that my wife is a gamesite administrator of eight years. Yeah, full disclosure. Yeah. So not only do I have a vested interest, it helps pay for these things. Sure. She had a concern, right? Like is this going to mean less job security for me, for my wife? So here's my hot take is no, it's not going to lead to issues for Gamesight administrators. And here's why. If you make it easier to build, you lower the bar to building and building more, then you put that into hands of people that don't understand the architecture and good design and the implications and the ramifications, then people like my wife will have to come in and clean up. Or if you're smart, you'll bring them in in the beginning to make sure that things are well designed. So that's exactly what I see. And I'll say this, there's not enough gamesite administrators. So the likely answer is that if you make it easier, you're going to put it in the hands of people that are underdeveloped in these areas, in these skills and experience, which will lead to more problems, which will lead to people like my wife having job security. That being said, I think the net net is still positive because if I had to choose between putting more resources and abilities in the hands of CS teams or less resources and abilities in the hands of CS teams, I'm going to choose more. Yes, that makes a lot of sense. Yeah, I agree with you completely. The sentiment that there aren't enough GAINSITE administrators in the world. And I think it might be a bit of an oversight sometimes because there's a metric ton of salesforce administrators, and granted, salesforce is much larger and you could say more complex, but the data structures are still. Yeah, no, I get it. No Gainsight, they're very well aware and they're trying to pull the same levers. Right. Universities and other programs to fill that. And I like Economics. It's an economics problem. You've got to convince people to take these courses to get these certifications, or you have to convince companies to foot that because then does Gainsight give it away for free? And then if they give it away for free, then it's a loaded cost with like it's a multi year wait to get there any sort of return on investment. It's a very complex problem. If you don't monetize your certifications in some way, then you're just going to have a bunch of people going through without skin in the game, as they like to say. Right? Yeah, exactly. Then, yeah, there's the whole value, price getting commitment. There's a lot of complex economical and behavior economical problems there, which I do find fascinating. But I realized this years ago. So I told my wife, I was like, you're going to be fine. They are not going to fix this problem overnight. No, I don't think so either. Yeah, that's cool. Okay, look, I think it's very clear in the CS community the kind of standard use cases that are emerging. You have things like quick meeting summaries based off of a transcript. You have account summaries from timeline entries. If we're talking about GAINSITE, we're talking about the admin side of things and being able to develop using generative AI. But I'm curious, what's your take on kind of some of the other use cases or edge cases or tools that you feel like are maybe undersung at the moment that we should be paying attention to as digital CS leaders? Yeah, I would say the term generative AI, though it's accurate, it can be a misnomer at times because the ability to summarize and analyze and extract is a very important thing. This is like the basis of update AI and our long term vision. We don't want to be a notetaker. That's not a good business to be in. We knew it'd be commoditized very fast. And guess what? That's exactly what happened. So the analogy that I was telling anybody who would listen at pools was number one. CS leaders are the only functional leader that do not have a proprietary data set. Everything that the CS leaders use is borrowed from other teams. Hence the C 360 design is literally just plug in all the different sources to try to create some picture of what's going on. Yeah. Where's all your data coming from? Pull it in one place. Yeah, but it's not yours. It belongs to other teams. That's a whole other host of problems. But the real problem is that if you're trying to have influence in an organization to drive change, influence, innovation, you need your own data set. The data set was actually already there. And the analogy is that your team has been having conversations on the phone that is a geyser of oil that has just been spouting out of the ground for years and you've been spending a shit ton of money. I don't know if we're allowed to curse, but I did anyway. You've been spending a shit ton of money to drill that hole. And then you take one little cup and you fill it up and you're like, I'm good, I got it. And there's all this oil on the ground you're not doing anything with. Yeah. Fill some fruit barrels. Yeah. So now there's a technology that can turn that oil into a myriad of different byproducts to not only help your team, but also your peers in other organizations. So when you're asking about use cases from a CS leader perspective, where they should get excited is I'm going to have data to make better decisions and to help my peers in other orgs, which means that not only am I going to have a better spot at the table, they're going to look to me to give them that data. And when the CFO comes around and says, we've got to divvy up this budget, people are going to think twice before immediately taking from CS, because then if you start to turn off that oil, those other teams are going to suffer just like we see time and time again with OPEC and actual oil markets. So that's what I'm most excited about for a bigger vision, is these tools ability to systematically extract insights from spoken conversations. Yeah. And turn that into something meaningful. Right. And actionable. Correct? Yeah. Because I think I was talking with somebody about health scores in general, and I think there's a common issue among scoring that's very reactive. It's like, okay, they've gone red. Why? Let's go look at see where the problem is versus predictive, which is to say, let's feed it all this stuff and really try to get ahead of any looming issues and things like that. And I think similarly to health scoring, I think AI can help be generative. AI can help be part of that kind of recipe of predictive intervention. Yeah, well, I mean, think about it this way. Health scores have almost entirely not captured conversational data. The way they usually are captured right now is a CSM has like a red, yellow, green that they can designate. I won't even get into the myriad of problems that that presents. But now we have a it's not objective, but it's it's a another source that is potentially more objective than a human that individuals, teams, leaders, organizations will be able to use to make better decisions. Yeah. And I know I try to be as platform agnostic in these conversations as possible, but we've been talking about GAINSITE the whole time. So GAINSITE has CX Center, which tries to pull out sentiment based on keywords that you feed it and whatnot. And I think that's obviously just scratching the surface of what we can do. Do you have any insight or thoughts on that? Like sending it from AI? Yeah, I do. Well, so we because we have a data, we have a head of data science here at Update AI, and we also have a woman who has a master's in linguistics. Because before we use Chat GPT in our product, we tried to build our own model. And what we found is that the English language is actually really tricky. It's not very scientific at times. Like, we tried to figure out action items. For instance, what's an action item versus what's not an action item? What's declarative versus presumptive? And then conditional? And long story short, it's very hard to do from a data science and statistical method using kind of standard regression analysis. So the good news is large language models are much better at it for all the underlying reasons. And right now it's an area where if you're trying to build health scores, if you're trying to understand sentiment, I think we have to rethink what that actually means because keywords, here's a really salient example of how they're fallible. For anybody that's married, you probably know that I'm fine, I'm fine, it's okay. That could actually mean I'm not fine. It's not okay. And guess what? Natural language processing, it doesn't know the difference. Right? It's black or white. Yeah, exactly. By the way, Chat GPT is the same. If you put it into Chat GPT and said Chat GPT, what is the positive, neutral, negative sentiment of this statement? It would probably say neutral or maybe positive. So tone, body language, all of those things not incorporated. That's why at Update, we actually don't have sentiment analysis in our product. We tested it. It wasn't actionable, it wasn't useful because we realized we're missing so much of what actually matters. So we think we could take a different approach. I'll give you a very specific example. If I asked you a question, a yes or no question, I would expect a direct response, right? Right. If you give me an ambiguous response and I don't pick up on that and I don't delve in, I don't probe deeper, and I kind of just be like, oh, cool. And I go on chat. GBT. Large language models, they could pick up on that and say no. Mickey asked a direct yes or no question, close ended question. Alex did not give a direct answer. He gave an ambiguous answer that could signal misalignment. And I think that that's more useful to the CSM, to the leader, to organizations than what's the sentiment? Yeah. It's almost like you're having to train these algorithms to do, like, leadership principles of, like, five wise the shit out of a question till you get the actual answer, right. Yeah. I'm assuming that people are already building or will build soon, like coaching, right. Where it's like, I know Gong has mentioned they're going to do live coaching in their platform. We'll see what it looks like and what they do. My guess is it won't be for customer success, unfortunately. Right. But I'm very much interested because I like to think I'm okay at it, but I would love to be a lot better to say, no, Mickey, don't keep going. Ask a deeper question. There's something behind that, and it'll be more interesting for you, too, right? From like, a podcast perspective. Sure. Absolutely. Yeah. That's fascinating. Well, I guess everybody is kind of in a wait and see mode. Like, we'll see where this is going. Yeah. It could be the end of the world. That's true. There's a distinct possibility. I saw the whole Tesla announcement of what was their robot called? Optimus or whatever. Yeah, something like that. Right. Got you. Okay. Yeah. Let's just build the robots. Go ahead. I think I saw a movie like this once. I do like that movie a lot. I do like terminator. It is good. Yeah, exactly. T two too. You can't forget t two. No. T two is my favorite. Yeah. Chance Cameron. Yes. Well, this is actually a pretty decent segue because I want to dive into some of the privacy concerns that the skeptics like to kind of bring up. I think there's a lot of companies right now that are publishing internal memos that say, hey, don't feed chat GPT, any proprietary information or personally identifiable information. That's like a non starter, and I get it. And it's cool to see, obviously, platforms adopting generative AI within their tools, which means that it makes the uptake of that palatable for companies because they're operating under privacy, blah, blah, blah, blah, blah. But I would love your kind of take on where the intersection of using generative AI models either publicly or within some sort of contained environment where the crossroads between that and privacy concerns are. Yeah. So I'll say, for the record, I'm not an It infosec, legal lawyer, anything that actually makes me an expert. But I have done a lot of reading on it, and from an update AI perspective, I talk with legal It infosec teams all the time because that is usually their first and last question. So, number one, this has always been a problem, by the way. You and I are communicating in a digital way. We're using Riverside FM, which I've used before, and it's great, but anything I say here is stored probably on AWS or Azure or Google. Right. So there's that. The second thing is OpenAI as of March 1, said that for their Chat GBT as well as their API, that unless you opt in, which is a very important distinction, unless you opt in, your information is not going to be used to train their models. So how much do you trust them? That's the next question. They've said that everybody has been impacted by data breaches or companies saying one thing, doing another. It's an industry. It's an industry. Yeah, it's an industry. Right. So that's kind of the first and last. Thing on that point. The second thing is there's already lots of development and there will be an entire industry around this. Meaning if I'm salesforce and I want to have large language models integrated everywhere in my product, but I have to keep them secure, well, I'll just pay $50 million to have my own instance that literally cannot be used elsewhere. It's trained on specific sets of data. I'm sure there's all sorts of safeguards. My guess is OpenAI is going to do like, that will be a major part of their business is you can buy a copy of GPT Four, train it, and fine tune it on all your own data. By the way, they're already kind of doing that with some of the companies that they've invested in. So, like, large language models as a service will likely be tens of billions of dollars in revenue per year business. So from a legal infosec leadership perspective, if you're okay with all of the current state, start using it right now, the risk is relatively low. And then in the long term, you probably will have much better security controls because, interesting, there's just too much money to be made. Yeah, absolutely. As you were talking, over the last decade or so, there's been this flood of going from On Prem to SaaS, and now with this, it's kind of like, well, are we going back to on Prem now? Might be. Yeah, it might be. I read a few years ago, Dropbox, they went from cloud and then they developed their own data center because they were like, wait, this is going to be way cheaper than the flexibility we get. So, yeah, I think it's possible. That not for everybody, right? Because a lot of startups need the flexibility, of course, but for some companies that can afford it. Yeah, they're like my own data centers, my own control, my own hardware. And my guess is there's an whole industry spinning up around that even. Can you buy plug and play data centers where you can just buy the entire data center and all the infrastructure is there for you, so you don't have to hire an army of people to manage it? Because again, we're talking about not enough people. There's probably not enough database engineers and all the roles that I have no idea what it takes. Yeah, you host or I host. All being hosted. That's right, it's hosted somewhere. Exactly. Hosted somewhere. Yeah, exactly. Cool. I feel like we could talk about this forever, but I've got just a couple of kind of wrap up things for you. The first is I'd love your take on I think you mentioned a book already, but do you have podcasts, YouTube channels? You're obviously a researcher, so I'd love to give everyone kind of a peek into what things you're using to research. Yeah, great question. So I did develop an ebook for Chat GPT in customer success. You can get a free copy of that by filling out an email form. Link in the bio. Yeah. Link in the description. That has a good amalgamation of kind of, like, how I feel, what the current state is, who I've learned from. There are some really amazing experts out there, like ali k. Miller, former head of machine learning business development at amazon, or ethan Molok, associate professor at wharton. If you're in education or, you know, folks in education, go read his stuff because he has mandated his students use it. Use chat GPT. So those are, you know, some of those are some of the folks you can find me on LinkedIn. That's the best place to find me and see what I'm posting and doing. Happy to share kind of what's coming up and what I'm thinking about as well. That's awesome. Yeah. I appreciate that. And who do you feel like is doing really stellar things in the realm of digital CS right now? Yeah, I mean, dan enos over@monday.com is obviously awesome. I've had lots of good conversations with him. I've generally been very impressed with notion and loom, and I know people at loom as well, and who else? Yeah, I mean, jeff at GitLab has generally been very impressive. They've done a lot of really great work over there. So, yeah, from a digital CS perspective, I think similar to chat GPT, we're very much in the early ages, like, the early stages of it, and I'm hoping maybe we can steal some more, like, marketing ops people or lifecycle marketing folks, because that's really what we need. It's a new skill set that it doesn't really readily exist within customer success. So we need more people coming to it. So I'm glad that you're bringing attention to it. We need more CS ops people. We need more people with marketing backgrounds, building systems behind it. I'm okay at it, but I've met people that are way better. So I'm hoping we can bring more of those folks into the fold. Yeah, somebody who's had experience with marketing automation is, like, worth their weight in gold in this stuff. Oh, yeah. Absolutely. Listening. If you have a background in anything marketing automation, there's way more jobs than there are than there are people to fill it. So if you want to stop competing for marketing ops jobs, we could use you. There you go. Yeah. Hit up mickey powell. Yeah, I don't know anybody specifically, but happily refer you to the people that I do know. Exactly. Cool. Hey, thanks for taking the time. I really appreciate you joining. It's been a pleasure chatting with you, and hopefully you'll come back at some point. Absolutely. Happy to do it, alex, thanks so much. Cool. Thanks,