Ep 33: Understanding AI and Customer Success with Gainsight CIO Karl Mosgofian

‍Saam Motamedi: Hi there, and welcome to Enterprise Software Innovators, a show where top technology executives share how they innovate at scale. In each episode, Enterprise CIOs share how they've applied exciting new technologies and what they've learned along the way. I'm Saam Motamedi, a general partner at Greylock Partners.

Evan Reiser: And I'm Evan Reiser, the CEO and founder of Abnormal Security. Today on the show, we're bringing you a conversation with Karl Mosgofian, CIO of Gainsight. Gainsight is an industry leading customer success platform, used by enterprise organizations to optimize the customer journey from beginning to end.

In this conversation, Karl shares his thoughts on balancing the hype versus reality of current AI capabilities, understanding opportunities presented by generative AI, and approaching AI integration as an enterprise team. 

So Karl, maybe to start, do you want to share like a brief overview of kind of like your career?

Kind of how did you end up, you know, uh, in, in this role of CIO of Gainsight? 

Karl Mosgofian: Yeah, I, uh, I've actually been doing this a long time, and I started as a computer hobbyist, uh, as a kid, and did all kinds of crazy things when I was still a teenager, and sort of fell into this as a career, starting more on the software side, but ultimately doing all aspects of IT.

I spent many years at Cadence Design Systems, doing a variety of things, rising up to a level where I was running the whole applications team, but also reinventing myself at one point as an enterprise architect, and going from managing a huge team to being kind of an individual contributor, but also having that chance to understand things at a deeper level.

And I think that that really positioned me well for the CIO role. So that then when I, when I had the opportunity to become a CIO, I was able to really understand all aspects of the technology landscape, but also because of my applications background, really understand the business side of things, right?

Because when you run things like SAP or CRM systems, you really have to learn a lot about how the business actually runs and business process. I think that's one of the things people who don't do enterprise applications don't always fully understand is how much that is a business process job and not a technology job.

And as time has gone by, it's become even less of a technology job. We used to have lots of programmers, uh, supporting those systems and a few business analysts. And now we have lots of business analysts and a few programmers. 

Evan: Maybe for our audience that's less familiar, do you mind just sharing a little bit about yeah, Gainsight, kind of what you do, and yeah, I'd love for our audience to better understand.

Karl: Yeah, I've had a great experience. I've been at Gainsight almost six years now, and we've really grown so much. We were really the pioneer in customer success software, and we've sort of grown up with the customer success function. As more and more companies have realized that if you're in any kind of recurring revenue business, if you just sell stuff and then leave people alone, you sometimes get some nasty surprises when it comes time for the renewal.

So you'd better have something, whether you call it a formal customer success department, or it's just a function that happens somewhere else in your company, you'd better have some way of staying in touch with your customers along the way, driving engagement, adoption, uh, and really what it comes down to is making sure that they get the value from you that they expected when they bought your product.

Saam: So as Evan alluded to, and Karl, the three of us were talking about before we hit the record button, I think it's hard for three folks like us to get together in November 2023 and have a conversation without bringing up AI.

AI is not a new thing. Yet, the last year has felt very, very different with the, you know, rise of ChatGPT, these generative models. Where are we in AI? Like, where do you see us in the hype cycle? Um, and what's your overall view on AI's potential for transforming enterprise? 

Karl: Yeah, I'm glad that you framed it that way because it's a little frustrating, I think, sometimes.

Some people are acting like AI was just invented, and it wasn't just invented. It's been around a long time. There's been a huge breakthrough, right? But we should have some perspective on that. For instance, Gainsight has had AI in our products for a long time. There are a lot of people out there who've been doing this for a long time, doing really effective stuff.

But it's also true that OpenAI made a big breakthrough. LLMs are really different. They're doing some things that previous technologies couldn't do, and it's super exciting, and it's going to have a big impact on the world. But the hype cycle is real. I've seen so many technologies over the years come along, and that cycle of people being super excited and then there being this backlash because, oh, this doesn't actually solve all the problems of the world, so this is terrible, to then finally reaching the steady state of, hey, it's a great technology, but it's got to be used in the right use cases.

That's always true. And the other thing that's always true is everybody always says, no, this one's different. So when it's at the peak of expectations, they're like, no, no, no, the old rules don't apply to this one. This one really is going to change the world. And the thing is, is that it's not to say that those technologies aren't hugely impactful.

I mean, the internet really did change the world in some pretty significant ways. At the same time, it's not like everything changed. One example of that that I reflect on is e-commerce. So, like, in the early 2000s, you had made some cool prediction, which was like, I can see the future, e-commerce is going to be huge, and it's going to take out retail. There's going to be no more retail. That actually would have been a really smart, cool take. And you could have written articles, and people would say, wow, that person's really a futurist, and it would have been a perfectly reasonable take, right? 

Well, here we sit 20 years later, and is e-commerce incredibly important and impactful in the world? Absolutely. Last stat I saw was it makes up something like 15 percent of retail. So is it huge? Yes, it's huge. Are there no more stores? No, there's still lots of stores. And, and I don't know, look, maybe a hundred years from now there will be no stores at all, and maybe that, maybe it's just a matter of time, but I think you can make a reasonable estimate around what's going to happen with the technology that is also wrong, or that is overhyped.

And so I feel like AI is kind of in that mode right now, where people are expecting it to do all things for all people. It's going to totally change everything. There are going to be no more jobs. There are going to be no more programmers. There's not going to be nothing, anything. And look, I don't know, maybe a hundred years from now that will be true.

But, we're definitely in the hype stage right now. 

Saam: So it's November 17th, 2023. ChatGPT came out November 30th, 2022. So we're just around a year. As you said, AI has been around for a long time. Like, did any of it surprise you? Like, is there anything over the last year that if I'd rewound to November 17th, 2022 and told you, Karl, this is going to happen, like you'd say, no, no way, like, has anything surprised you on the upside in terms of what's been demonstrated with AI?

Karl: So, I think I've been hugely surprised at the leap, right, because, you know, GPT is so much better, so much faster than I would have expected. And it went from this sort of crude, basic tool that you could sort of see some glimmers of, to being able to do unbelievable things really easily. And so, while I'm sort of realistic, about how much of an impact it can have, where and when and how.

That doesn't take away from the fact that it's an incredible technology. It makes me think of the Arthur C. Clarke line, that any sufficiently advanced technology is indistinguishable from magic. And ChatGPT is maybe the most magical technology I have ever seen. So, in some ways I'm all in on the hype, in the sense that this is really an incredible technology that's capable of fantastic things.

And by the way, we're at ChatGPT 4, right? So what's ChatGPT 17 going to look like? Running on quantum computing. I mean, if this is what it is in the early stages, what's it going to look like in 5-10 years, right? So I'm super excited about it. And I think it's adding a lot of value for a lot of people right now.

So the summarization use case is beautiful, right? Just give it a prompt, feed it some data and say, clean this up, right? Um, I'm using it all the time. You know, in the IT world, there's a lot of documentation or communication that's written by technical folks. So I'll get something and it'll be a page long of technical stuff and I'll just feed it to ChatGPT and say, can you rewrite this for a non technical audience?

And it does a really good job. It's not perfect. It's, you know, I think that's one thing that people kind of miss sometimes, which is that this technology is not something that just automatically does things and you don't have to read the output. It's just good. You know, you have to tweak it. You still have to put some energy into it, but it's amazingly good at that. 

Um, a good example of that, by the way, is that we added a feature to our product. So we, we're all about the customer 360, right? So we collect a ton of information and, but it's almost too much information, right? So what we realized was we could shove all of that into an LLM and just have it summarize it and essentially give what we call a customer cheat sheet. Right. 

If you think about it, everybody has the situation where the, the CEO says, Hey, I'm gonna go meet with this customer. Can someone just give me the few bullet points? And that's actually the hard part, right? In some ways, if he said, give me all your information, it'd be easy. We could cut and paste a huge amount of information.

It's summarizing it that requires a lot of human intelligence and time. And, and that's what this is fantastic at, right? So we were able to, to create this feature and it works really well. Um, you know, and, and What's amazing about that too, by the way, is like, you don't need data scientists to implement that feature, right?

You don't need to be an AI expert to run your communication through ChatGPT and ask it to rewrite it for you as bullet points or something. Like anybody can do that. So it's also super democratized. So that stuff is surprising to me that we so quickly got to a point where this very sophisticated technology is so consumable by so many people so quickly.

Evan: Are there kind of like upcoming use cases that you think there's going to be potential to kind of drive impact against that in terms of improving how you operate the business? 

Karl: Absolutely. And, you know, we're pursuing a ton of different projects to try different things. So there's a lot of support use cases. Either for external support or internal helpdesk kind of stuff, but that's just like a natural use case for this technology. So we're definitely pursuing that in our support organization and within IT, um, to see where that can help us, right? 

These GPT enabled chatbots, I think, are going to just be Much better than the previous generation. Having said that, we've had chatbots for a long time. So if chatbots were meant that you didn't have to have support people anymore, that would have already happened. And so the question now is, how much of a productivity increase is a much better chatbot going to give you? Is it going to be enough to make a significant impact?

Or is it going to be a nice productivity boost, but You know, but not necessarily something that's going to totally change the face of customer support. And that's where I'm, I'm a little bit more cautious in sort of over promising. And I think the danger there is sometimes people look at that kind of use case from a distance and they're like, Oh, great. We'll get rid of the whole support department. We'll replace it with a chat bot. 

What I've seen so far, we're not there. Maybe chatbot version 47 will be able to do that. But, you know, with where we're at now, I don't really see that happening. And it's also still very dependent on knowledge and information that those human beings need to provide to it.

It's super interesting. It's super exciting. We're experimenting. We're seeing what we can get out of it. But we're trying to be a little bit cautious about overpromising what the business impact is going to be. 

Evan: Are there other areas where you, you worry that kind of like the industry more broadly is kind of overestimating, you know, some of the, the capabilities, whether the efficacy of the speed to achieve those, or maybe, you know, where they're kind of underestimating some of the challenges or obstacles that are going to get in the way of the vision versus the reality.

Karl: Yeah, I mean, I guess I sort of alluded to it, but everything always comes down to data, right? So you can have this fantastic tool for synthesizing data, but it's only as good as the data. And so that's one of the areas where I think people sometimes get tripped up. They do a POC and it looks pretty good, but then they try and implement at scale, and they realize that their data is not actually in good enough shape to give them very good results.

So I think one of the things that AI is going to do is force people to go back then and say, okay, this will do all sorts of great stuff if we can clean up some of these data issues, right? We can have this chatbot that works super well, but we need to give it a base of knowledge base articles to work off of and we don't have those. So we need to go create those. 

So data is one sticking point, but I think the main thing is just very easy to overestimate, I think, how much of a productivity gain you can get from some of these tools. Again, based on just playing around a little bit or reading some article that's kind of at 50, 000 feet. 

You know, the programming use case, I think, is another interesting one. I think it's easy to look at this from a distance and say, oh, we won't need programmers because I can go into ChatGPT and say, write me a script that will do this, and it writes it. And so you say, oh, so this thing can write programs for me, and the new programming language will be English, and that's it.

I do think this will evolve and it will become more and more useful. But the thing is, I don't think you can extrapolate from that to, I'm just going to tell it to go write me a customer success platform, and it's going to go write it. And I think to be honest, I think that sometimes misunderstands what programmers really do too.

So if I go to ChatGPT and say, write me a bubble sort algorithm, and it writes it for me, that's nice, but that didn't replace somebody sitting and writing that, that replaced somebody going to a repository and cutting and pasting some existing code. So that's great. I mean, it's, it's a, it's helpful, but it's not like it's, changes everything about how programs are written.

So, you know, I think that there's great productivity gains, but It's not this sort of automatic thing that's going to completely eliminate whole functions The way that a lot of the the articles and stuff would would make you think, so that's the area where I feel like we Really have to kind of set expectations with our internal leadership with our boards of directors with people who are, kind of looking at this from a distance like great, how are you gonna take cost out of my, out of my organization?

I think ultimately we will take some cost out of the organization, and we will create more efficiency and we'll take some grunt work off of people's plate, which is always good, right? But in terms of it completely changing the whole paradigm and we know, you know, that I don't think we're there yet.

Evan: And Karl, one thing you said that I just think you said it really well is if you're detached from the work that's actually getting done, you may not be clear on what the role really is. You know, being a support rep is not just about pulling technical questions and answers out of like the knowledge base.

ChatGPT can totally do that, right? Part of it is understanding the context of the customer, reading the emotional cues and the question, right? Trying to kind of give people, um, assurance and safety and a clear path to resolution, making sure they feel like they've been heard, right? Like, there's a lot more to that job than just like, hey, can you just find this for me in the knowledge base, right?

And so I think that, It's a good lesson for all of us, like, let's make sure that we're not oversimplifying what we think some of these jobs are right. And I'm assuming I will do everything, but there's certainly some known parts, right? Whether that could be optimized or improved. But yeah, there's a reason why it takes talented people to do this work.

Karl: Yeah, it's an interesting dynamic and, and, you know, I've experienced it my whole, my whole career. You know, one of the interesting things that happens is that everybody understands their space. So if you're in accounting, you really understand accounting and you appreciate the complexity and what people really do.

People outside that team think you just push numbers around. They think what you do is probably a lot easier than it really is. But that's true of everybody, so everybody's like, my thing's really hard, but why do you need so many people? Why can't you just? Those are the words that drive me crazy more than anything else, but that's what it comes down to, right?

There's all these people in the world saying, why can't you just? It seems so simple from a distance. Why can't you just do this and it would be easy? And there's always a real good answer. I'm like, no, it's really complicated. And this is the thing. This is the essence of kind of the struggle that I think IT departments have a lot, inside of a business, right? Because we're trying to prove value and people like, well, why does this stuff even matter? But it's because there's a lot of complexity in the world and people don't always appreciate it. And so, yeah, I think the way you said that is exactly right,

that AI just exacerbates that and makes people think, well, if I can ask it this question, get this answer, then surely it can do the job of accounting. Cause their job is easy and you know, right. Except that none of it is easy, right? 

If it was easy, the thing is I've been doing this stuff a long time. If it was easy to automate, it would have been automated. And we've automated a ton, right? So all these business processes that used to be super manual are now highly automated, but it didn't mean we don't need any people at all. Like we still need people because it's still complex. 

Saam: Yeah, I agree with you, Karl, and I, it's, as you were speaking, I was thinking about some of these demos I see on Twitter where it's like, you know, build an entire end to end application using, um, using, uh, Copilot, and to your point, like, that's a very small part of, like, what actually goes into engineering, and so those things make for great demos, but they're likely not going to manifest in that exact way, um, in how like real enterprises do software development. That's not to say there won't be big impact and change. And so maybe speaking or like thinking about what some of the changes could be. Like what, like whether you look at your right to organization or you know, we have a lot of CIOs and CTOs listening to this podcast who are thinking about their organizations. What are the changes or impacts one should think about organizationally? To take advantage of these advancements. So, so like, are there any new roles that are going to emerge? I don't, you know, you can imagine something like a GPT knowledge manager who's responsible for, for, for the, you know, the knowledge that goes into GPTs.

I'm just thinking out loud, but what are the impacts? 

Karl: So one of the first things I did was identify someone on my team and said, I want you to be the AI person. I want you to be a focal point and run a sort of center of excellence within my team, just to start, so that somebody is really on point. The danger of things like this, especially when they're so easy to, like anybody can go to ChatGPT, is like, you've got all these different people in the organization, they're doing cool stuff, but nobody's talking to each other. There's no sort of security or governance around it. There's no, it's just like the wild west. And a little bit of that is okay. Right. I mean, I don't want to like stop that and somehow try and control everything, but I actually really think that in the modern world, the role of it more and more is actually not to be controllers or, sort of dictators, but to be facilitators and coordinators across the company. And so what we quickly evolved to was actually a broader cross company group, which has representatives from all the teams and there are regular meetings, there's a slack channel, there's a place for us to collaborate with each other and say, Hey, what's everybody doing?

Right? Let's learn from each other. Right? Like I tried this use case and it worked really well. Oh, great. We have a similar kind of situation. That's good to know. Or, we tried this and actually it was really hard. Has anyone figured this out? So, we were very active in helping create that collaborative work within the company without necessarily saying, well, everything has to come through us, or we have to run everything related to AI.

I'm definitely asking everybody on my team, what is your AI roadmap? And we're really working with each of the business groups to say, what is your AI roadmap? Everybody's got to figure out how this plugs into your world. Now, in some cases, maybe it's light, right? Maybe it's, hey, I run my communications through ChatGPT and clean them up before I send them out.

In some cases, it may be huge. It may be, well, we're really going to use AI chatbots to really significantly change the support experience, but everybody's got to have a roadmap and a plan. So, you know, we've definitely tried to support the efforts of the whole company around that, but certainly in my team, it's a huge focus.

Evan: One of the struggles I hear from a lot of, you know, CIOs I work with is this balance between, on one hand, you're trying to like drive adoption of innovation in these new technologies, and that involves, Not just new technology, but even new like habits and workflows and people's day to day job. 

At the same time, right, you're also making sure people don't dump into like random online websites like all your proprietary information. That's actually people aren't overly trusting these outputs. And so how have you approached building a culture that kind of drives innovation? Right? You mentioned like your team working different business functions to go innovate and think through that road map.

So like, how do you, um, how do you kind of get people to really innovate right without kind of Putting unnecessary obstacles in place, but also making sure you have the right guard rails so that you, you can drive fast. 

Karl: Yeah, I think I guess there's there's two pieces of that. I think one is it's really important to have partners in IT who work with the business teams, not as sort of order takers, but as true partners.

And in some ways, you can think about it as being like a CSM. So in the same way that customer success does that with your external customers, we need to think like a customer success organization internally. And how are we enabling our customers to be successful with technology? 

But I think the other thing that, and this is something that's really different than it used to be, I think tools like Slack have really changed the game in terms of democratizing how we work together. So look, we're a startup. We've always had a very federated environment. IT doesn't run everything. A lot of people go off and do their own applications and things within their groups, and that's okay.

And the approach that I've always taken is, I just want to know about everything. I want to be like the enterprise architect. I want to be the glue that's helping hold everything together. We do data warehouse, business intelligence, we do data integration. We do those things that are sort of horizontals that cut across everybody, but we don't try and dictate every little thing that everybody does.

But we do want to be able to be that clearinghouse, that place where it all comes together. And, obviously, we also need to secure everything. So that's the less fun part, but, like, an important part. So we also need to know everything so we can make sure it's secure. But the interesting thing, as I reflect on my long career in IT, and how things used to be when it was much more centralized, and IT played a different kind of role, is there's a lot of just people in Slack saying, Hey, I'm looking for a tool, does someone have something that does this. Or, the way that we communicate the way I communicate to the company is often much less formal, right? It's not like here's an official communication from I. T. And thou shalt do this or that shall do that. It's more. Hey, folks, zoom has this new feature where it'll do an AI companion. Check it out. It's really cool. Right? But be careful. Read the output. Don't just cut and paste it because you might have to edit it a little. It's not always perfect. Um, and I think about that now, and I think that's so different than like 10 years ago, and, and how I used to communicate to my users, and how my users communicated to me. I just think it's much more open now, and I, and I like it.

It creates some challenges, right? I mean, there were some things that were nice about when we just ran everything and nobody could do anything without talking to me. And I had control of everything, but on the whole, I like it the way it is now better.

And I, and I think it's much better for the company from an innovation standpoint, right? Because we really aren't creating a lot of barriers to people innovating. There's a few guardrails, right, for security and things to make sure we don't get ourselves in too much trouble. But in general, I think we have a very open, collaborative culture where people feel comfortable innovating. But we also keep from stepping on each other's toes too much. 

I think that's the other thing when you talk about that coordination is somebody does something in one department, it works great for them, it breaks something for someone else. You know, we're trying to avoid those kinds of situations. And the bigger you get, the more important that is because everything's so interconnected.

Evan: We have like two minutes left. So, um, Karl, you got, you got to give us like the one tweet version. So when we hit our true lightning rounds, so the sound, you want to kick it off real quick? 

Saam: Yes. Karl, how should companies measure the success of a CIO? 

Karl: Success in helping the business meet its objectives. 

Evan: What is one piece of advice you wish someone told you when you first became CIO?

Karl: Be bolder than you have been in the past. Take a chance. Don't worry so much about getting in trouble or stepping on toes. Have an opinion. Go for it. 

Saam: Switching to the personal side. What's a recent book you've read that's had a big impact on you and why?

Karl: So a very old book, but that had a huge impact, and that I find myself thinking about a lot is Gödel, Escher and Bach. And the interesting thing to me that as old as that is, it's still highly relevant to what's happening in AI now. And it gave me some concepts and a sort of framework to think about AI that I find is still useful. 

Evan: All right. And Karl, final question, but it's gotta be like the two, the three word version or less.

What's an upcoming technology inside work or outside of work that you're most excited about? 

Karl: So, something that I really have resisted getting excited about is the virtual reality headset stuff. But I have to admit, I'm starting to kind of get sucked in, and I'm pretty curious to see what this Apple thing actually looks like.

Evan: Karl, thank you so much for taking time to chat with us. Looking forward to chatting again soon. 

Saam: Thanks, Karl. 

Karl: Thanks, this was great. 

Evan: That was Karl Mosgofian, CIO of Gainsight. 

Saam: Thanks for listening to the Enterprise Software Innovators Podcast. I'm Saam Motamedi, a general partner at Greylock Partners. 

Evan: And I'm Evan Reiser, the CEO and founder of Abnormal Security.

Please be sure to subscribe so you never miss an episode. You can find more great lessons from technology leaders and other enterprise software experts at enterprisesoftware.blog. 

Saam: This show is produced by Luke Reiser and Josh Meer. See you next time.