In a special episode of Behind the Deal live from Thoma Bravo’s Annual Meeting in March 2024, Managing Partner Seth Boro welcomes Sumit Dhawan, CEO at Proofpoint, Mike Capone, CEO at Qlik, and Charlie Gottdiener, CEO at Anaplan, to the stage to discuss how their companies are leveraging the power of artificial intelligence. The CEOs share perspectives on generative AI and its impact on cybersecurity, what AI means for the future of IP and how businesses can channel AI into actionable solutions for their customers.
AIR DATE:
April 25, 2024
LENGTH:
47:48
Transcript
HOLDEN SPAHT:
Hello, and welcome to a very special live recording of Thoma Bravo's podcast, Behind the Deal. I'm Thoma Bravo Managing Partner, Holden Spaht.
We're coming off an amazing second season of Behind the Deal.
SCOTT CRABILL:
Well, if the auto companies want you to be bigger than you are, you've done that service because you brought $300 million in revenue when we bought JD Power. Today, we're approaching $900 million. How'd you pull that off?
DAVID HABIGER:
Well, you guys were helpful. Thank you, uh, for the advice and capital.
HOLDEN SPAHT:
We initially created this podcast to give an inside look into the transactions and partnerships that shape our portfolio. Over the last year, our partners sat down with our portfolio company's CEOs to explore the stories and lessons behind some of our most dynamic software deals.
JIM RYAN:
Orlando never flinched. He got into his car, drove south down the 101 and [inaudible 00:00:51] the deal from this, this other private equity firm. The rest was, was history after that.
SETH BORO:
I remember that very well. And, uh, it was an exhilarating evening of deal making. And it was also the first step in that partnership.
JIM RYAN:
Yeah.
HOLDEN SPAHT:
But perhaps the best deal we made was with our listeners. Behind the Deal has received over 500,000 downloads and topped the US business charts on Apple Podcasts.
ORLANDO BRAVO:
Firms change, firms evolve, partners leave, there are spin outs, people start new firms. And at Thoma Bravo, we've been at this together now, for over 25 years, nothing has changed. We are still here doing the same thing. And frankly, as we say it all the time, I feel we're only getting started.
HOLDEN SPAHT:
Thank you for listening. We appreciate your support. And now, it is my pleasure to introduce Thoma Bravo Managing Partner, my friend, and more importantly, Behind the Deal cohost, Seth Boro.
CARL THOMA:
This is Behind the Deal, Live From Miami.
SETH BORO:
Well, that, that was, uh, that was awesome. I'm, uh, a little nervous here. I have not done a live video podcast before, uh, in this format, but, um, I think we're gonna be okay.
We have a, we have an awesome, awesome topic today that I know a lot of people have a lot of interest in, um, because we get this question constantly about AI and, and what's going on in our portfolio.
Um, here today, joining me, um, from three of our flagship companies in Fund XIV and Fund XV, are Sumit Dhawan from Proofpoint, Mike Capone from Qlik, and Charlie Gottdiener from Anaplan. It's a nice representation of applications, uh, with Anaplan security and, and, uh, in the cyber world, the email cyber world with Proofpoint. And then, uh, data and, uh, and analytics with, uh, with Qlik.
M- Maybe, um, just to level set, um, you know, AI now, is in the vernacular, um, constantly. It's, you know, something that a couple of years ago, uh, of course, we know about but we weren't talking about it as much. Um, and it would be great just to get each of your perspectives, um, you know, uh, in terms of, you know, how we got here.
So, why is it today, that we're, you know, talking about it? Um, it's, it's front of mind, it's hard to read a, a business article or a company article, uh, without talking or thinking about AI. What, what, what's changed? Uh, Sumit, maybe we'll start with you.
SUMIT DHAWAN:
Yeah. Good afternoon, everyone. Um, listen, uh, I sort of say, Proofpoint has always had AI. We did c- cybersecurity, you can't have people defending all these other attackers, you have to use technology. And the technology needs to keep learning. So, by definition that's classical machine learning.
So, I jokingly say, Proof, Proofpoint did AI before AI was cool, and I call it sort of the, the before cool era, or BC era of AI. Um, so we, we have done all forms of machine learning and AI models to essentially prevent attackers that are constantly attacking through email and other forms of communications that people have. And AI is the technology or the model as the way you beat all of... You, you do pattern detection. And then you, you detect this is a potential attack. And you don't let it go to the, to the employee or the end user as we call it in the tech jargon.
Um, and now, what's happening is with the generative AI, um, basically it's consumerized where everyone can experience the power of AI because you can actually talk to it. It can generate responses, images, all forms of m- m- models that are available in terms of how you experience AI. So, it's readily available, and it's readily available for both good guys like us to use. As well as in the world of cyber, the bad guys where, you know, AI can be used to potentially create new forms attacks. And AI can be used now, for preventing those new forms of attacks, so.
But in the world of cyber, we are certainly seeing more and more, um, how this generative AI, both in terms of threats as well as how you prevent the threats to become more and more relevant and important going forward.
SETH BORO:
Mike, what, um, you know, you, you've obviously, um, in Click, been delivering insights to your customers in various ways over time. Um, how has this shift, uh, to generative, um, enabled you to deliver more value? And I, I'd actually be curious also, Charlie, to, to hear from you when Mike is done in terms of what kinda business value you're able to generate today with this, with this major platform shift that, um, that's taken place, really over the, over the last year.
MIKE CAPONE:
Right, right. Well, Sumit is right. First of all, this is not new. I mean, we, we are writing, um, AI programming when I was in college, you know, some 20, 30 years ago. Um, this, this kind of breakthrough into consumer with ChatGPT and LLMs, um, has been abled by the massive computer power that's out there today. It's available to us in cloud computing and just more sophistication in kind of the techniques.
Um, we at Qlik, we've been getting ready for this moment for the last five years. We were, um, a analytics company. And yeah, we had AI built into our platform. We've been doing NLP, natural language processing, in our platform. We've been able to interact with data using modern AI techniques for a long time. But what's changed is, with this massive compute is the ability, uh, for companies now, to harness all of the information at their fingertips.
And this kind of explosion of AI and this kinda renaissance that happened last year, was a huge wakeup call to companies to suddenly realize, "I have to build a foundation to take advantage of AI." You can't just plug in an LLM or ChatGPT. At the end, you gotta do the work.
And it's what we see in these companies really, um, start to scramble to build an infrastructure and a foundation that secures and governs their data for AI, and make sure that you don't run into kinda problems that you're seeing out there today, um, around false positives, hallucinations and things.
And so, the series of acquisitions that we did including leading up through Talend, um, was really to get ready for this moment. So, we've getting ready for this moment for the last five years to be able to harness data for enterprises to actually effectively use AI as part of a corporate structure, versus, you know, consumer, you know, writing high school term papers, for example, is not what we're, we're trying to do.
CHARLIE GOTTDIENER:
Yeah. So, um, before I talk about the value points, I just wanna add one thing. Agree with both of w- everything that both of you said. I think the other thing that's important to recognize is that AI and ML was really the domain of data science experts, right? And that expertise really isn't needed as much as, it's still needed to, to make advances, but it's not needed, needed to engage in AI like it once was because it has been consumerized by the LLMs and, and the o- the open AI architecture.
So, I think that was also a breakthrough, right? Where you just don't need the data science expertise to just engage with AI that we did two, three years ago.
Um, now, back to Anaplan. So, we think about creating value for customers with AI in three different ways. Uh, the first is driving more insight for them. So, Anaplan, you know, what customers do on Anaplan is they forecast and plan their businesses. And what AI, and we've had AI products now for over five years, what AI allows them to do is to make more accurate forecasts because we're, we're running ML al- algorithms because it's mostly machine language, uh, uh, models across much larger data sets. Much deeper data sets, much larger data sets, both first party and third party data. And that improves forecasting. So, that's kinda the first vector of value.
The second is access. So, this is where we get into generative, right? We're, we're really, um, allowing customers at all different levels of the organization to access their models in a much simpler way. So today, primarily, people who access e- their Anaplan models, are model builders. They're experts that have to go into the platform and dig out the insights. It's not hard to dig them out, but you have to know something about the platform.
Tomorrow, we're gonna leverage e- we're, or really, we have a prototype now, where people are gonna be able to ask their models questions using natural language. So, a CFO could ask for a forecast, right, on his way to work, and get that answer back without having go, to go through an inter- intermediary that's expert in the platform. So, that's, that's today. And then that's actually getting tied to workflow, so they could actually ask, send the, the question to somebody as a followup. So, that's really second point of value.
The third is really efficiency. So, we think about developing a copilot for model building. So, t- you have, again, you have to be an expert to build a model in Anaplan. That takes a lot of time. You start from scratch. And what the copilot will do is it will, it will, it will build a model for you. And then model builders become editors as opposed to creators. That's, that's much more, uh, productive and efficient.
SETH BORO:
Sumit, in, in cyber, there is, um, so many threat vectors and, um, you know, clearly, uh, the advent of generative AI has created yet another one.
SUMIT DHAWAN:
Mm-hmm.
SETH BORO:
How are your customers, um, managing those threats today? Maybe walk through an example of what you've seen in the field, just in terms of some of the more modern day threats that are coming out of the, the use of, as you said, the, you know, the, the bad guys using generative AI, of course, it helps, you know, that groups as well, and it's very well funded.
And then what are you doing on the flip side, uh, to help combat some of these new threats? And how quickly have you had to react to that?
SUMIT DHAWAN:
Yeah. Uh, firstly, I think every one of you has probably experienced an email or a text message that comes in, either trying to make you click on something, or sometimes even some, something coming from your CFO or CEO for, you know, a gift card that you wanna buy. It's pretty common attacks, both for fraud as well as for, uh, basically, you know, planting a little malware into your network or your computer, which can then eventually become ransomware attacks. And you know, m- many of your firms have either been through it, or, you know, always prevent from doing it.
It turns out, more than three-quarters of those attacks start from email, and they continue to be that way. Uh, so if you think about cyber, cyber always started from network, you know. Hey, I'll prevent my computer network and make sure bad guys can't get in. But social engineering which is people attacking people, is the biggest surface area for most attacks coming in.
And that's what we focus on. We have three trillion emails that we, that go through our system, about one and a half trillion SMS messages that go through our system, which give us a mechanism to build models. Models that can predict which is a good email and what's a bad email. So, at any given point of time, we are able to see as a company, probably the only company in the world who can really see ahead of the curve on what is the emergence of new types of attacks, which is what, Seth, you were asking.
So, using all these data, we have a threat research and threat intelligence team that continually s- continuously publish, publishes to our customers and the world, how the attacks are evolving. What we are seeing now, is that generative AI at this point of time, is being used more and more by attackers, which we can tell. We can tell when an attack which is coming in through an email or a spurious looking website that people are, uh, that, you know, users that are being, uh, used to click on because they look familiar, they're all being created using generative AI. So, in other words, the same attacks but are being created by generative AI.
What are the, w- Uh, so what does that mean? That means, oftentimes, these attacks are created with non-native language speakers, that's no longer an issue. So, we can't just sort of make these emails and websites through language assessment if they are threats or not, that's no longer the case. We have to throw that out. And we have to use other indicators to tell if it's a threat or not.
Secondly, there is more context you can build. Oftentimes now, generative AI based attacks are not going to be where they send one email to a bunch of people as a campaign. Instead, because they are being robotically generated, generative AI enables effectively a robot to have a conversation with several individuals in this room, which is more contextual to you as individuals, which makes the models that we have to have to protect against those kinds of threats even more sophisticated.
It needs to have more information on, if I'm, as a sender, sending you those kinds of emails and trying to induce you to click on something, then using our technology now, which is also based on these large language models, we can detect. You know what? Sumit never really sends an email to Orlando asking for credit card numbers, or you know... (laughing)
SETH BORO:
Mm-hmm, mm-hmm. Just once.
SUMIT DHAWAN:
I mean, I, I, uh... Just once.
SETH BORO:
Mm-hmm.
SUMIT DHAWAN:
Um, but, um, but, uh, but... So, so that would basically clear the contextual model to say, "This sender is not the r- This is not the right context for this type of email." And that's the kind of sophistication both threat actors are using and we are evolving in our models to make sure we can protect. And that's a good thing in my assessment for incumbents such as us because I start this by saying we have 3 trillion emails a year that we process, 1.3 trillion, you know, SMS, 17 trillion URLs that we see.
What does that mean? That means we have the data. We can train these models better than anyone else. It's hard for an incumbent like us to be disrupted by someone because building a model is not necessarily the hard, uh, building the tech and the code for these model is not the hard thing, training that with the right model is the hard thing. And that's what incumbents benefit from.
SETH BORO:
Cl- Clearly, um, the importance of data has, is something that we've talked about for a while, but it f- it feels like we're now in a new realm of, of, um, the value and owning that. And Mike, I know you're helping, you're enab- you're enabling customers in this journey. Can you help us understand, um, what your customers are asking for today as they get ready to use these large language models to create value for their customers or internally? And what does that look like at Qlik? Um, are customers ready today? You know, do they have budget? Do they have the talent? H- What, what does that, you know, what does that process look like right now?
MIKE CAPONE:
Yeah, sure. So you know, a year ago, we were all staring down this explosion and there was almost a panic out in the market, right, as. And boards were yelling at CEOs, saying, "Do something, do anything." And people are now, saying, "Initiatives, spending money." But really, it was just to show that they were, you know, paying attention and, and not, um, and, you know, not being afraid of being disrupted.
The good news is, a lot of that has settled down right now. Um, the hype has sort of settled. And now, we're into this phase of people being a lot more thoughtful, which is great. And what they're saying is, "I need to build a foundation. I need to build the right foundation, um, for my future of AI." And really, it comes down to three things.
First is, how can I get the maximum value? The second is, how can I govern it and be secure, so I don't land in, in jail? You saw the EU just pass some really sweeping legislation, um, recently, um, around what can happen if you, um, you know, if you misuse AI. And then third is cost, right? There's, there's cost to c- associated with building out these infrastructures.
Um, on the data side, what, what we've been building, what we do, is really simple. Which is, data from anywhere, any source, um, cloud, on-premise, uh, wherever, at high velocity in realtime into kinda modern cloud data link structures, but also, governed, cataloged, um, with veracity, proven data lineage, um, all in one place. So, you can govern it and aware, not only where your data is, but where it came from. And then you can analyze it using modern kinda techniques, like analytics and AI and natural language processing.
And then something Charlie said before is really important, you can action it. So, once you get the insight 'cause the CEOs will tell me, "All right, I got the answer. The analytics and the AI gave me the insight, but now, my company doesn't do anything with it," right? So, the ability to actually take that and put it into a workflow, put it into an automation, send an alert, um, integrate to an RPA platform, super critical. Otherwise, you don't get the value. Um, that's what happens.
So, that, that's kinda the, the holistic view of things. Um, and then what I would say is, um, on the, on the cost side of things, you know, people are starting to get the first anniversary of their, their cloud data link build, right? They're starting to understand what's happening and how much money they're spending to move data around, and they're, and they're, they're freaking out, right? They're saying, "Well, I can't, this isn't sustainable, if I wanna harness all of my data."
So, what we do is we built a series of, um, analytics apps using the power of our analytics platform on top of our data platform. It's how customers manage, putting data in the most cost-effective place to still drive the right outcome, but make sure that, you know, when you're paying for compute, you're paying for, um, data movement in your storage, that you're putting things in the most cost optimal place, and still get in the AI outcome, that's really how people are thinking about it.
And the good news is it's moving at a more thoughtful pace now. It's not just do anything to say that you're doing something. Um, there are budgets. People are spending. Um, a lot of it is top down, CEOs saying, "You earmark a lot of money, uh, to get on AI," but now, that money is being deployed very, very thoughtfully.
SETH BORO:
Charlie, obviously, um, you need internal talent to, you know, work with all these new technologies. I, I don't know if you consider this a platform shift or an enabling technology. But maybe help us understand, um, you know, what that has looked like as you're running the company. And in terms of what sort of internal resources you need to actually turn this into a product that's revenue generating for you. And, um, has that been a challenge given how quickly it's moved? Or, or how, how have you managed that so far?
CHARLIE GOTTDIENER:
Yeah. So, I'll go back to something I talked about earlier. So you know, this all started with really a data science sort of academic experiment, right? That's where machine learning really started. It became more commercial and practical over time. As we get harder, there's more to compute, larger data sets. And so, the shift that we've seen in talent is assessing whether our data scientist can make that shift to be more commercial, to drive commercial outcomes from their, uh, machine learning models and really drive AI that's gonna drive the business.
So, I talked earlier about, you know, us, uh, leveraging generative in, you know, I've been told we have named the product, ASK Anaplan, so the CFO could actually, um, uh, query models through natural language. That's a harder shift for, uh, academic data scientists. They, they don't think about the world that way 'cause they wanna run the next experiment. They wanna build the next, uh, you know, large language model that may or may not have a practical application.
So, that's really our challenge, right, is kinda figuring out who can really make the leap to commercial because we're not, we're not an academic institution, right? We wanna leverage the knowledge of academic institutions and we wanna commercialize it.
So, that's, that's, that's the journey that we're on.
SETH BORO:
Okay.
CHARLIE GOTTDIENER:
And we've got a fairly big team that's trying to make that journey, that's, uh, in Israel, they came through an acquisition for us.
SETH BORO:
Mike, you mentioned the, um, EU legislation that came out last week. Maybe walk people through, 'cause you're, you know very deep in that. You've, you have formed a c- uh, an A, and AI council around governance and ethics. I think it would be great just to get our perspective. And then Sumit and Charlie, you s- your perspective on, just from a governance and ethics perspective, um, how this is being managed internally at, at your companies 'cause this, it is a big issue.
MIKE CAPONE:
It's, it's, uh, it's a huge issue. And, and just like every other technology, um, the innovation is gonna outpace society's ability to kinda regulate it, right? It's just, it's just how the world works in, in, in today's world. Um, and so, um, we, in the corporate world, actually have an even greater obligation to govern, to govern this, and to make sure that what we're deploying, the tools we're deploying and the technology, um, is used in the right way. And can be governed in a way that you can, um, you can control the, the kind of the ethics of it.
Um, and so, we're really thoughtful about that at Qlik. And, um, you know, for the most part, I'd say, most, most companies in our industry are behaving that way. Some, some aren't, but most are. Um, and so, um, you know, if you think about... My, my favorite kinda paradigm is if, if you think about social media, you know, 20 years ago and how that started taking off, and what that has done to kinda teenagers in today's world, would we have approached that the same way? Would we have let that go the way it did, completely ungoverned for so long? Or would we have been more thoughtful about it? I think that's where we are with AI today.
And so, we at Qlik, um, spend a lot of time thinking about our responsibility, um, uh, to the ethics, um, and the morality of AI. And what we, what we know is that we, we don't know. Like, we just, we haven't figured everything out. So, we formed a council of people that mo- mostly outside of industry, people from public sector, people from academia. Um, there are, are four very respected individuals, come from, like, Cambridge and things like that, who actually advise us and our customers on, um, the ethical deployment of AI. And how we can do this, um, in the most, you know, uh, capture all the values.
Certainly, we're in the valley of creation business. But also, do it in a thoughtful way, where, uh, people view us as a, a trusted partner that helps us, that helps them be successful the, the right way rather than the wrong way.
The EU legislation is a great example where, um, they've really, um, if you read it, they've gone quite deep. And they've categorized different uses of AI. Um, some of it being completely prohibited. So, using AI to predict criminal behavior, even to predict ethnicity or gender, um, is some things you might not be able to do anymore in, in Europe with your platform, if this really takes hold. Um, and they've categorized different ways and it puts a huge burden on companies to make sure they're using AI the right way and governing their data the right way.
And like, you know, the EU, they never do anything, like, light, right, the EU Parliament. So, the penalty is up to, um, 7% of turnover, 7% of revenue if you screw it up, right? So, you really need to be very thoughtful and very reminiscent of GDPR. And so, you know, we have to help customers. All of us have to help customers make sure, um, that they're doing this, um, not only fast, um, and [inaudible 00:23:44] our value, but the right way, morally, ethically, and in compliance with the law.
SUMIT DHAWAN:
Yeah. I think, uh, given, with Proofpoint, I mentioned, we have data from customer data. So, one of the big things in addition to obviously, h- our models being trained the right way and are doing the right things that are compliant. We also have to be extremely careful with data, data residency laws that are there. Uh, so data residency laws are different across the globe. And they become even more important and critical when you're applying these models on them.
Um, and there are, you know, it becomes quite complex. Like, for example, uh, I, I think everyone here has heard of, or maybe even familiar with or experienced, Copilots from Microsoft. Um, one of the banks that maybe we have folks, uh, here as well, they, I heard from them that they were doing a pilot and Copilot was linked into certain business applications as well as e- I don't know if it's, it was Copilot or some other generative AI chat bot that was prepared, that was connecting through different business applications. And they were running a pilot. And they enabled a set of users to start doing queries.
And once you s- enable business unit users to have that form of freeform access to information that's in multiple business applications, that type of queries that people start making can expose the information that sometimes regulated and protected in certain ways, which you know, you never, um, thought anyone would sort of be able to access. So, all of a sudden, it is opening up a whole new dimension of data, security, protection, governance, requirements for customers.
And so, I do see in terms of adoption of the technology f- even from the customer end, where they are all going through this sort of phase of how they are gonna tackle cost and governance elements. And for our end, we have to be much more thoughtful in terms of what we do with our own data governance for all kinds of regul- re- regulations that exist from sovereignty perspective, residency perspective, or in general compliance because of the new, new rules that are coming in.
CHARLIE GOTTDIENER:
Yeah, I, I agree with that. I, I think, um, many of you may be asking, so what now, right? What do companies do? Um, I, I think the news is actually a little better than it was, you know, 5, 10 years ago with, from, from a data privacy perspective. Because there are regulations and there are models that companies have, have, have followed for data privacy and, and, and data governance.
Um, and so, there are techniques like privacy by design that are, that really govern the use of data in product innovation in most companies. Not every company is great at it, but at least these kinda concepts exist. And I suspect AI will leverage those to begin with, but will need, as Mike, you point out, a lot more governance because it can be a free for all. And privacy by design is actually a pretty rigid process that forces data governance roles. But if AI is in everybody's hands, then it would be hard to, to really manage that and police it, so.
But I do think the, the news is a little better, there's a starting point. And most companies have, uh, data privacy and, and data governance, um, committees, setups, to, to, as a starting point. And that's where I, where I see, uh, a lot of companies starting, is in those committees.
SETH BORO:
There's, um, there's, there is, you know, you, you hear talk about the fact that generative AI could be an issue as it relates to IP. And you know, the, the competitive landscape within, uh, certain industries, especially within software, where maybe there's an ability to spin up a new company quickly. Can you share your thoughts on, uh, you know, on, on that? It... How, how big a risk is it to your businesses? What should we all be thinking about, um, you know, as a, as investors in software as it, as it relates to that?
MAYBE Sumit.
SUMIT DHAWAN:
Yeah. I mean, first of all, I think, uh, as I mentioned in the world of cyber, broadly speaking, there is infrastructure and human level protection. We focus on the latter. That's the biggest surface area of risk or threats. And the generative AI is actually, to some extent, a bit of a tailwind for us because it increases the potential of attacks. Makes the effective, makes, uh, creating an effective attack easier and more available and accessible to bad actors at lower cost, than them trying to do it themselves, okay.
So, we are seeing that as potentially a tailwind. And as long as we have the right technology to defend against those types of attacks, in general that sort of, uh, grows the need for a sort of robust solution like, like, uh, like Proofpoint.
Secondly, as I mentioned, at the end of the day, you need more sophisticated models to prevent against those types of attacks. And those models can only be, sure, you can have, uh, coders write code faster, but without access to that data, you're not gonna be able to develop those models.
And so, there is a inherent incumbency benefit. Can you get data elsewhere? Maybe, possibly, but nowhere the same quality of data that we have as an incumbent when we are already serving enterprise customers and delivering their critical asset, like, mail and, you know, other data protection solutions. So, I think second is that it, it does give benefit to incumbents.
I'd say, third, um, I wouldn't, I don't think we are sitting still, right? We are leveraging the power of generative AI and these large language models to create more upsell and cross-sell opportunities. For example, you know, you may have all experienced it. You may have fat-fingered yoursel- uh, y- you know, a wrong, wrong person and sent s- you know, not just an attack, but just information that was s- not meant to be sent to someone, sent, that you sent an email to. And then you sort of sent a note, "Oh, can you please delete it. That was not for you." It's happened to all of us.
Uh, we have built, uh, because we are in the mail flow, um, we've built a solution and we already have the context, the example that I gave, where I sent that attack to Orlando. Instead of an attack, it may have been just an email that I accidentally was sending to, uh, to Orlando. And our technology now, would prompt me to say, "Hey, do you really intend to send it to Orlando? This doesn't look like, you know, the type of information you usually share to him."
So, it's a nudge as a, as AI technology. And that's extremely helpful. If you can think about, even in your businesses, in large banks, there are teams of 40 to 50 people that are simply checking outbound email to all sorts of clients, if that email is the right one to be sent or not. We can take all of that cost out and create that incremental monetization all because of the power of gen AI. So, enabling new use cases because of the technology.
So, so, A, it's a tailwind for us in the cyber. B, incumbency helps us and doesn't hurt us in any ways. And C, we're not standing still. We're building new monetization and new solutions that are serving real problems and, and, uh, uh, for, for our customers, which are much, much easier for us to bring to market than anyone else.
SETH BORO:
Charlie, do you, do you worry about somebody building an Anaplan competitor, uh, quicker and easier n- you know, with, now that code is easier to generate and you don't need to be a specialist anymore?
CHARLIE GOTTDIENER:
I, uh, um, I don't. Primarily, because if I'm exposing my customers' data to public generative AI models, I'm done, right? Because they now, own my customers' data. And so, we have a proprietary platform that really protects that data and the usage of that data. So it... To me, a lot of this comes back to data at, at its core, right? Generative AI is gonna a- allow us to get a lot more insight out of data. And so, you know, uh, what we're working on, you know, to, to advance the insight dimension, um, that I talked about earlier, is really leveraging generative AI, uh, to, uh, do more dynamic forecasting. So, really early warning systems, right?
So, you can imagine a company, let's say it's a manufacturer and they've got distribution, they've got big retail customers, think Walmart or Target, uh, really big customers. And now, they've got external events that they actually wanna do, put into a, a w- early warning system like weather, right, would be the simple example.
So, if my big customers are running a massive promotion on my product and I've got a weather event, I could pretty out of sync with where inventory is in my distribution centers, and I could, I could, I could a- really, uh, risk, you know, uh, irritating them 'cause I, I have, they have a lot of stockouts when they've got a big promotion coming, right?
So, we're gonna use AI to actually create early warning systems, and that was a simple example, across multiple data sets, so we can help our customers forecast even more accurately, and actually get ahead. Really get into the predictive business, so they can avoid stockouts in my example, um, a- and, rather than, you know, l- what are we gonna do about a forecast when we know we're gonna miss inventory levels, right?
So, that's kinda where we're headed, uh, with generative AI.
SETH BORO:
Um, we've, we talked a lot about our, our portfolio today, uh, and driving efficiency. And you know, currently running at a 40% margin, sometimes more. It would be great if you could walk through examples internally. So, we talked about your product. We talked about what you're doing to customers, the market. Um, how can you leverage this next generation of AI technology to, across your organization, Mike, maybe we'll start with you, um, to drive operational efficiencies, um, in all aspects? Or, and w- And where do you see it today, and where do you see it going?
MIKE CAPONE:
Yeah. You know, today, today, I would, I would call it, the low hanging fruit exercises. So, where we get a lot of lift, so for example, code generation, right? We're, at the end of the day, we write software. Um, the ability to actually, uh, generate code that's fairly, you know, standard, very usable, is with AI has just gotten much, much better. We've gotten so far beyond, you know, kinda open source to now, where those, those tools are publicly available. And, and that's how we're able to run our R&D percentage lower than a lot of our competitors, right, in additional to low cost locations.
Um, you know, the other area we're having a big impact right now is in sort of the sales and marketing area, where we're able to do a lot more targeted, uh, selling, targeted marketing, um, using our own capabilities inside of Qlik. And then use some third party tools like 6Sense, et cetera, are actually looking at kinda buying patterns. Who's in the market? Let's target them. Um, make sure, you know, our, our worst enemy is, is not knowing, right? Uh, data and, uh, BI deals come and go before we see them.
Now, with kind of the tools out there, we can look out, in the market, look who's doing certain searches and predict who's about to likely buy our software, our competitors' software and actually go target that. So, that'll be, I think that's the, you know, that's the, I, I don't wanna say the easiest stuff, 'cause none of it is easy, but it's, it's right in front of us, right? It's, it, it's right there.
And then, you know, going forward, I think, um, you know, for us, you know, we're a pretty simple business, right? We buy software and we sell software. But the, our opportunity to help our customers who've got massive amounts data, who've got a lot of, um, you know, I'd say, kinda muddy processes and, you know, just trying to fight through stuff, is gonna be, um, is gonna be almost unlimited.
And, and the, the real, uh, the, the real kinda differentiation we have, um, I'm really excited about the Talend business because, you know, a lot of people think that having more data is better. And more is not always better. Sometimes better is better, right?
So, the ability to actually take the data that a customer has. We were talking earlier, Seth, about a large automotive manufacturer who built their own infrastructure, their own LLM, their own, um, their own ecosystem. And what we're helping them do now, is harness all of the data inside of their four walls, um, including the data that comes off the cars and the sensors and everything, in realtime, and make decisions about inventory, about, um, about where to put the next charging station, for example. Um, and that's, that's, like, a huge productivity, uh, lift for them.
And then to protect IP, nothing goes out. And when they bring data from the outside, they can, uh, prove the veracity they have, they can curate that data and make sure that it's not gonna inject, um, bad data or bias into their data set. So, for me, that's the most exciting part of all of this, is what we can do for all of our customers.
SETH BORO:
Sumit, it would be great to hear about how you, today, you admit 30s margin. How can, you know, what, what do you see operationally, internally, to move that to, I don't know mid 40s, is it 50%?
MIKE CAPONE:
We're being recorded, just so you know.
SETH BORO:
Yeah. It's a commitment.
SUMIT DHAWAN:
It's a commitment, yeah.
SETH BORO:
Yeah. What, [inaudible 00:36:46].
SUMIT DHAWAN:
I think, I think, first of all, we are, we are, we are in a similar state, starting with, you can call it low hanging fruit. Although, the results of those have been astonishingly positive, uh, in the code development. Uh, just developers have embraced it. The success that we are seeing in terms of code that comes in as a suggestion that people accept, which basically is another proxy to show that there is productivity intrinsically being improved in terms of how fast we can build new codes. So, that's certainly there. Uh...
SETH BORO:
Sumit, how, how are you managing the security on that code-
SUMIT DHAWAN:
Um...
SETH BORO:
... which I know is a big issue?
SUMIT DHAWAN:
No. We, uh... Same, same thing as the, uh, these are, uh, these are sort of, you know, where the code is, um... You know, code right now, sort of goes into your own instance of GitHub largely. So, these are appropriately sort of governed private instances of, uh, of code m- that is running. Uh, so the right security for IP protection is being put in place. But I do, we do think that it's improved productivity without any compromise of IP leakage, that's been done.
On the, I think the, the one we are, where I'm most excited about where we are just starting to do, is more predictive analytics on churn because we have a lot of indicators of information from customers, on usage of the product. And how through that, we can create patterns and very quickly then have the right reactive, or quickly corrective measures in a more proactive m- manner put in place.
Uh, today, a lot of that is done somewhat manual, somewhat reactive. And m- And just, you know, the, you can imagine resource allocation is not sort of really been put in place as proactively as we could. So, using the s- m- Using models, you can think of them as still fairly simplistic AI models. But theses days, those models can greatly improve your gross retention rate. And all of a sudden, just a single point of gr- gross retention rate can give a significant boost to the earnings, um, just, uh, flows through.
So, I think, um, I think that's, uh, that's the one, in addition to sales and marketing, I'm excited about, we're starting on that. And then there is some education, tech support, again, public information that's available. How we can just streamline it, so that there are less calls coming in, and improves the productivity of when we actually serve the customers calling in. We are doing so in a faster way, um, and, and whatnot.
Because usually there's a lot of time taken by tech support engineers to train them and whatnot. And these days, no one really wants to call. So, if there is an agent that can serve, uh, a large volume of support, support requests, that's the, another investment we are making to make sure we can be more efficient.
SETH BORO:
Great. Charlie, and anything that you're seeing in your business? Or...
CHARLIE GOTTDIENER:
So, so, so very similar themed. So, I, I won't re- repeat what these guys said. But w- One thing I'll add is we're, we're starting to leverage AI pretty deeply in sales productivity. And that's an area that we're working on improving in the business. It should give us some real margin expansion. And we've leveraged a third party, uh, AI tool to do this.
But one of the things that we've discovered is that, um, we went into last year thinking that you had to have eight meetings a week to reach your quota as a salesperson. Uh, that activity level would drive it. What we learned was that, it's actually not eight meeting. It's actually four meetings with director, uh, level or above that generates actually 125% of quota achievement.
And so then the question becomes, well, who has the relationships, right? What does the relationship map look like? And of course we find that we're not where we need to be, to generate those four high quality meetings a week across somebody's portfolio.
And so, that, uh, allows us then, then to ask the next question, which is, how are we gonna leverage AI to access the right relationships, whether it's third party data via LinkedIn or something else? And that's really what we're working on now, because we know that that relationship exists. That, you know, better meetings with more senior people allows us to focus the activity and be more productive at the same time.
Um, and, you know, when we marry that with, uh, a product that we actually have, an AI product called, Predictive Insights, and what the product does is it helps, uh, customers identify the ideal customer profile. So now, you've got the right customer profile, you've got much more targeted meetings, and that should be a much more productive formula for driving new bookings.
So, all of that leverage is AI.
SETH BORO:
Um, m- maybe with the last question, how are you balancing where we're at today, and, um, the pace of innovation just in this area going forward? And, uh, you know, how, how much time are you spending on the future, versus the present? Because I think this is, this, maybe this platform shift is moving. I don't know if it's a platform shift or this gener- this, you know, the technology innovation here is moving as fast as anything we've seen. And the absorption rate has been really high. And the fact that you're using it today and we're talking about it in a way that's very tangible, um, is impressive given we're really just a couple of years into this generative AI phase.
But Sumit, how, how, how do you think about that and manage that, just that pace of innovation? And any thoughts on what's next?
SUMIT DHAWAN:
Yeah. I think, first of all, this technology as many of you have pr- probably read in terms of just adoption from global population, all that, it's off the charts in terms of ChatGPT adoptions. So, in terms, for us, we have to be as a cyber defense company, a step ahead of the bad actors. At this point in time, the internal machinery at Proofpoint is assuming that the, the, the, the bad actors have access, freeform access to this technology. The threats will be created using this. And we are leveraging the technology to its fullest when it comes to building the defense against it.
There is no... Th- That's a topic, our, our CTO office is, is, uh, is sort of squarely responsible for it. Uh, we did an acquisition of the company called, Tessian, where a lot of effort was, is going into integrate. It brought in a large language model of our own, um, with the right sort of cost structure that can build a very rich contextual graph. You may have heard this term, which essentially creates a relationship between a large number of objects, so that you can very quickly, you know, identify, uh, patterns that are off the, you know, the norm.
And so, um, so we... So, both through organic efforts as well as our investments via M&A, we are, on that front, we are, you know, we're not assuming this is something that, that's gonna happen in the future. We have to stay a step ahead. And that sort of gives us credibility with our customers that creates momentum with our customers. It drives somewhat of growth as well for us. So, that's sort of our assumption, uh, when it comes to our core business, uh, cyber.
For, for areas that we, we do, we have data protection, uh, portfolio and business as well, which is our growth business in addition to email security. Our data protection business, we, we are closely monitoring their, what tangible innovations we could build, so that when customers start adopting AI for their critical lines of business, uh, application use cases, the example that I was giving, where there was a chatbot talking to various applications. How our data protection engine can just, how we can ride that wave as that adoption happens?
So, for defense solutions, we are assuming it's here. For our data protection solutions, we are very closely monitoring how and when the customers adopt it, so that our innovations are not behind. But, but it's, it's a, it's a topic we're discussing on an ongoing basis. We can't, uh, we can't afford not to.
MIKE CAPONE:
I think, for us, um, you know, we, we built a lot of structure to make sure we're that, um, we're in tune, right? So, I talked about our AI council, which is really important, right? It gives us one perspective. We have a, a large customer council, uh, that advises us on AI and what they're doing and what they're seeing in the market. And we have representation from all industries on that, that council to help us go.
I'll tell you though, the, the biggest secret weapon we've got actually is our M&A strategy and our partnership with Thoma Bravo. You know, we've got, you know, Mike Hoffmann and, and Mohnish. Um, and we're constantly scanning the market, looking for cool AI companies, what they're doing. Because a lot of times, the best source of innovation is to look at what's out there in the market, where's the venture money flowing, and what's going on out there to predict kinda what the next wave of innovation is going to be.
And that is super helpful in kinda understanding what we need to get ready for, what we, what we should be buying. And, you know, as we see, you know, dries up in some of these opportunities, we're gonna be all over them, um, trying to, to buy, you know, pennies on the dollar for what went into these things.
CHARLIE GOTTDIENER:
Yeah. And I would say, you know, we're, we're going at a measured pace, right? We're kinda earlier in the whole than I think, uh, some other companies. We've just gone through a big transformation or the early stages of transformation. So, we're, we're gonna, uh, continue to take our AI, uh, legacy, leverage it into the products and the platform that we have.
I, I can imagine a future, and I've started talking to my, uh, head of product and technology about this, where we have a leader of AI products. Um, and that will, you know, help us with an, with a, uh, an m- an M&A strategy to, to augment what we do there. And that's a conversation which we've just started. But I, I suspect we'll get there some time this year 'cause it is such a big topic, and it's a big topic for innovation.
SETH BORO:
Um, well, we are... The shock clock is almost out. We... This has been great. Um, I think the thing that's most interesting is, oftentimes, you know, you, you hear a lot about in the venture industry, innovation taking place in, in areas like AI. One of the big takeaways today that we would like everyone here to understand, is that our companies are innovating in and around these areas in big ways with massive R&D budgets. Not only delivering, um, innovation to customers and, uh, into the industry, also, creating policy, you know, thinking about ethics and governance, which is incredibly important. But then also taking this new technology and driving it into internal operations to produce better business results.
So, thank you for the great work. Uh, this has been fascinating. And, uh, we, we really appreciate your time and you being here today.
MIKE CAPONE:
Thank you.
CHARLIE GOTTDIENER:
It's our pleasure.
ORLANDO BRAVO:
This is our final episode of season two. And I wanted to take a moment to say thank you to all of you for coming along on this journey with us. We've gotten to share some incredible stories about our partnerships and we appreciate all who listen.
We'll be taking a brief break to record more episodes, but we'll be back again to take you Behind the Deal later this year. I'm Orlando Bravo. Thanks for listening.
Disclaimer:
Certain statements about Thoma Bravo made by portfolio company executives are intended to illustrate Thoma Bravo's business relationship with such persons rather than Thoma Bravo's capabilities or expertise with respect to investment advisory services. Portfolio company executives were not compensated in connection with their podcast participation, although they generally received compensation and investment opportunities in connection with their portfolio company roles, and in certain cases are also owners of portfolio company securities and/or investors in Thoma Bravo funds. Such compensation and investments subject podcast participants to potential conflicts of interest.
