Into the SingleVerse: An Inside Look at Insight with Lonne Jaffe & Teddie Wardi

My guests this week are Lonnie Jaffe and Teddy Vardy, Managing Directors at Insight Partners. Lonnie began his tech journey at the ripe old age of four with a Commodore VIC-20. He honed his programming skills at Stuyvesant High School and Harvard while interning at Microsoft. Lonnie joined Insight Partners in 2017 and focuses on AI, machine learning, healthcare software, fintech and other innovative technologies. Teddy is a veteran venture investor focused on high growth software and internet companies. Teddy also joined Insight Partners in 2017 where he works closely with portfolio companies and leads new investments. Before becoming an investor, Teddy co-founded and served as CTO of NovoGrid, a cloud services marketplace acquired by Also Group. Lonnie, Teddy, welcome to the podcast. Introduce yourself a little bit about your journey in tech. What got you started? Lonnie, of course, we know you got started at four with interest in programming and how it led you to tech. And yeah, and then we go on a wandering journey from there. I actually had a fairly similar journey. I started with a Commodore C64, so slightly different platform, but I think quite a similar trajectory as Lonnie. He was five or six years old at that and liked to tinker with software and building software. I was coding basic when I was like eight years old and went from there. I ended up founding a startup when I was still undergrad, spending seven long years on the journey, learning a lot about the ins and outs of building a business, but also about the kind of software and hardware stack and things like how the internet really works and so forth. It was like the pre-cloud era when you actually had to have servers, physical sun servers in Iraq. So it was a more physical approach to building companies than it might be today when you just harness the cloud. And then after exiting that, I moved to the venture investing dark side and have been investing in application software and infrastructure and cybersecurity for the past decade. And so this is Lonnie. So I'm one of the MDs at Insight. I joined about six and a half years ago. It was May of 2017. I used to be CEO of an Insight portfolio company though. So that company was called Syncsort at the time. It's now called Precisely. It's an older company. It was founded in 1968. So before I was born, I did not start it. And I spent about 15 years at IBM and CA Technologies before that, both also fairly old companies that were founded before I was born. And yeah, at Insight, I've been focused on some of the tech domains that we invest in. So things like AI, which we'll get a chance to talk about. That's great. So you and I have now, Insight invested in single store in 2020. I remember October, November timeframe. And it was an interesting journey to get to that place and has been a very rewarding journey since. So if I haven't done that before, which I know I have, thank you all for your support and your investment. It literally saved the company and we wouldn't have been here without you guys' support. It's been really, truly wonderful to work with both of you. Now, I've been an operator all my life and I don't really know what it is to be an investor. You both have been operators and now investors. So what do you think are the top two or three differences between your life as an operator versus your life as an investor? Maybe Lonnie, you go first this time and then Teddy. There's a lot that people will talk about, about the differences. One somewhat maybe contrarian perspective on this is there are elements about it that are maybe not as different as people will sometimes think. So as an investor, you're usually working with more companies, more legal entities at least. So there's separate C-corps or LLCs and you don't necessarily invest and end up owning all of each of them. You can invest in a company like single store and own a certain percentage of that and you can get a control position. At Insight Partners, we do venture buyouts where we can take control of a company. So you can end up with governance control, but maybe still not 100%. And you have extremely high caliber CEOs like you who are leading the companies. And it's not exactly an employer-employee relationship. It's a little different, but there are some similarities there. And you share management responsibility to some degree with the other investors and board members who are involved, which is not a thing that usually happens in companies. Usually the org chart, everybody only works for one person or they may have certain dotted line relationships. But when you actually look at the day-to-day, there can actually be a lot of similarities. You're helping the companies become more successful. If you add up the revenue of all of the portfolio companies in a place like Insight, it's like a large software company, like a really large one, not totally dissimilar to what you would see at the IBM software business or something like that. And there are some significant advantages of the investor form factor. People who work in a company can get equity in the thing that they're working on, the actual smaller unit. Whereas in a larger company, you have to end up designing more operating metric type incentive structures to give people economics in their day-to-day because often the parent company is just so big that it's hard for any single person other than the CEO and some of the senior execs to really be able to affect the outcome of it. And it can also be a little bit more elegant configuration-wise. You can invest in something, you can make it worth a lot more, and it's a little easier to exit it. You can keep it for a long time. And we have really long-term time horizons with our investments. You don't have to keep it forever. A large multinational corporation will often keep things and only divest of them or spin them out if they're not doing well. Non-core, but often that means it's going to be a euphemism for not doing well. And so that can be a positive. But of course there are differences. The pacing is very different. I spent some time doing M&A at a couple of different larger companies, and we're able to do really high-caliber diligence with just a lot more urgency. So we're able to get deals done much more quickly. And we don't have quite as much infrastructure at the fund level as you might get at a large multinational publicly-traded C-corp, let's say, but we do try to replicate a lot of it. So we have a team called Insight Onsite, which at this point is over 120 full-time people, and they're designed to be functional. So you have sales, marketing, product management, technology, architecture. Within sales, you have people who work on sales rep, territory design, and recruiting, and comp planning, and the marketing center of excellence. You have people who do category creation workshops, or who can work on building out account-based marketing systems, or manage relationships with Gartner and Forrester. So we do try to recreate a little bit of what you would get from being a business unit within the context of a large multinational corporation, because of our scale. There's a lot of advantages to scale, like the ability to amortize expensive fixed costs over a portfolio. Teddy? Yeah, I would say that the key difference is just being an operator, being an investor, it's really about a single bet versus a portfolio. And how that manifests itself, I would say, is that when I was an operator, I would spend all my time with my team, like management team employees, and my customers. And sometimes you spend some time with your investors, and then sometimes you see your competition at a trade show or something a couple of times a year. And it's a very focused way of doing things, and you're doing this one thing 24-7. And then as an investor, you work with multiple companies, five, 10, 15, whatever is standard in the investor world. So you're going to have this portfolio where you're context switching constantly. And then the result really is that when you're an operator, the highs are very high, but then also the lows are very low. I remember when I was an operator, when we won a big contract, it was one of the best moments. Raj, I'm sure you're going to relate to this, winning a big deal, making a quarter, that's really great. But then the lows are also very low, like something bad happens to the business, like you turn a deal or you have an attrition in the management team that you don't like. These things feel pretty dramatic when it happens. And then as an investor, somebody making their quarter or not making their quarter, it's just the lows are not as low, but the highs are not as high. And it's just a different approach to it. And I think that's the reason why sometimes some operators are just private operating that they might try investing and they find it less exciting or it doesn't really provide the adrenaline rush that they like in operating. And that's why sometimes the personality types are a little bit different. So I think that's really the key difference. It's like a single bet versus portfolio and how vested you are in the success of that one company versus many. Yeah, well said. The thing that I find really amusing, really, is when you get on the internet, you say, check out the deck that helped Raj raise $100 million or Peter raise $50 million. And they have this, they almost make it sound like there is a secret sauce in the deck that gets investors to invest in people, which I know is not true. Of course, you know, it's not true. So let's talk to our audience about, you see a number of companies coming your way at various stages for investment. What are probably the top two or three mistakes that CEOs and management teams make when they come to investors like yourself? How about Teddy, you go first on this one. Yeah, I think the key one is that if you look at the pitch, really, is that the pitch has to be tied to the stage of the business and what you're trying to raise. Often people say the series A's are raised really on a vision or some like, well, Cedron is raised on a vision and team. Series A's are raised on some proof of traction, but most of the vision, and then series B and growth rounds are raised on proof and metrics and so forth. So I think sometimes founders go a bit wrong in the fact that the way they pitch their company and kind of what they highlight and what they try to prove isn't really a match. So you might be trying to raise a series C, but you're actually talking only about some like future vision, like five years from now, whereas the growth stage investors only want to see that page 18 with those metrics and KPIs and retention and growth. Another way around, you might be raising a C drone and you're showcasing just the fact that, okay, you actually have revenue, you post a 100K contract, whereas people are investing, is this going to be the next IPO? They actually want to hear the story. So that would be one thing that there's a little mismatch in kind of what you highlight for the stage of investment. Lonnie? Yeah. So one of the nice things about growth stage investing is in a way, the world does some of the work for you of understanding the business. You've sort of offloaded the task of figuring out if things are working to the world. There's the growth rate and the revenue and things like that, but the revenue itself has a lot of, I'll call them quality ingredients. So things like high gross margins, high recurring revenue with high customer retention rates, rapid expansion within customers, things like that. And you can see improving unit economics, decreasing tax over time. And these are all not guarantees, but they're strong signals of the sustainability of the business and customer's love of the product and the potential for future cash flows. And if you have other things about the business that will make it durable, and what I mean by durable, I mean is it can keep growing in an efficient way for a really long period of time. Like a company grows at 50% a year for 10 years, it's 50 times larger, 50% a year for 20 years, and it's what, like 3,000 times larger. So it makes a very big difference to have a business that's very durable. And that happens, right? You see companies like Google and others that have been able to grow at those kinds of pieces for that length of time. I'd say a common mistake is when a founder or a CEO tries to obfuscate all of that information. You'll see things like putting the details of how the business is doing traction-wise towards the very, very end, or to just cherry pick certain metrics to show and kind of leave out others. And no business is perfect, right? Everybody will agree on what the platonic ideal of really high-quality business might look like. And what you're disagreeing about is which things that are not perfect you're willing to accept as risks or challenges. And you just need to see all of that upfront, right? Obviously having those things be better is better. And then being really effective at articulating them. And then I would say there's another thing, which is just being able to tell the story about the business in a way that's compelling and evocative, right? What about this is the reason why it's going to be sustainably differentiated? What kinds of things improve with scale? There's a lot of aspects of businesses that get better as the business grows. Things like network effects or data businesses will often be able to aggregate more data, and that's a benefit. There's huge supply-side economies of scale that get better as more people use the product, things like that. And just being able to tell the story around why the product itself will continue to be compelling over time and maintain competitive advantage. Lonnie, so you mean it's not enough to have a two-by-two with the competitors where your company is always up on the right to knock on the door, right? Exactly. How much or what quality, which is not quantifiable, do you look for in a founder or a CEO that is sitting metaphorically across the table from you, both that will basically be a, well, we aren't investing in because of this, or we will invest in it because of this? I mean, the qualitative stuff is quite hard, but it's also very important. I think that's why investing throughout the pandemic was quite difficult as you had to make investment decisions purely within Zoom meetings instead of meeting people in person. And at least I believe that meeting people in person and spending more time than at one hour Zoom over time, you just kind of learn much more about people. At least what I look for in CEOs is a person who can kind of work through various situations that arise over an investment journey. When you invest in early growth stage in these companies, it's often a, in successful scenarios, it could be like a seven or eight-year journey together. And the companies grow to like 10, 20, 50X the size, both in terms of revenues and headcounts and so forth. So it's really about figuring out if the person is someone who can kind of stay with the journey for a long time and kind of adapt him or herself to dealing with all these situations that we see arising in software businesses over time. So it's quite non-scientific in a way, but you're really kind of like testing like, is this a relationship that you see as a long-term workable thing and you see as a person who you can kind of work through some tougher situations, not just the kind of good board reports and kind of fantastic outcomes. Yeah. And then there are some things that, you know, the CEO is a manager. And so there's just good leadership and management things that you will often want to see, right? The ability to make good decisions quickly to hold the organization to high standards. One thing that people are sometimes concerned about with less experienced CEOs, you know, sometimes technical founders who haven't run a sizable company before, are they going to be able to hold people to be accountable for performing well? Are they going to be a pushover? But, you know, they also want people like really good people want their CEO to be kind and fair too. They don't want them to be mean or aggressive. And it's critical to be able to recruit and motivate teams. So it's probably one of the most important skillsets and being the kind of person that people want to work for. And then I would just say, there's like a lot of things that can go wrong day to day in a company and sort of having the temperament to get energized by those kinds of challenges as opposed to frustrated by that, right? Because you're constantly needing to work through unglamorous tasks that get in your way and, you know, and figure out creative solutions to problems and things like that. And, you know, some people really enjoy that and some people find it really not pleasant. Yeah, I know. In fact, a lot of times people ask me, you know, what was my experience becoming a CEO? Because as you know, I was COO for a long time. And I often wondered, you know, how much different could, you know, number one be compared to number two? And I articulated and I said, you know, I became funnier overnight and lonelier overnight to become a CEO. It is a very lonely job. Like, that's what people are going to underappreciate. Yeah. And actually, that was the biggest surprise. And intellectually, you get it. You get it. Yeah, it's going to be lonely. There's that and the other. And once you... One of the things about being an investor that's actually kind of interesting is you get to work with a lot of CEOs, whereas like as a CEO, you don't. I mean, you can like personally try to network with them, but it isn't as much as part of the day to day. You actually learn a lot about CEO-ness, you know, in the investor role where you're dealing with really high caliber CEOs. Yeah. In fact, one of the most cathartic things that I tend to do is I've got a foursome at my golf club and the other three are CEOs as well. And it's just it's just cathartic just to speak to them about some of the issues that we have and just make fun of each other's companies and whatever is happening. But you're right. It's a very, very lonely job. And I never... You know, that was, I guess, the biggest surprise for me. 2021, the madness in the investment world. In fact, I met one of my old... In fact, I sold one of my companies to him and he has now moved to the VC side of the pair. And he looked pretty suicidal. And he's like, I have invested 60% of my portfolio in 2021. So it's been... It's not been a good year because of whatever, you know, the multiples having dropped, etc. So how would you articulate what we on the other side of the fence or the public at large calls it the 2021 madness where companies were getting just unreal multiples, etc. Lonnie, how do you, if there is a word like rationalize it, how do you think about it? Yeah, I mean, when we invest, we're typically underwriting a business case that involves multiples coming down over time in pretty much any market environment. And the reason for that is as the companies grow, their growth rate does tend to slow. Not always, but it usually tends to slow down. And as a result, the multiples tend to just compress. When multiples are higher, like higher than normal because of the sort of macro fluctuations in multiples, you want to underwrite, choose more significant multiple declines. And this may seem sort of obvious, but it's useful to say some of these things out loud. Like if a multiple, a revenue multiple, let's say for a high growth company comes down significantly, let's say it drops by 50%. But the company is growing at 100% a year, which is not uncommon for high growth businesses. And the multiple doesn't continue to come down, right? You don't have to wait that long before the absolute value of the business is well above where it was, and then it continues to grow at 100% per year at that point. The real challenge, I think, coming out of 2021, I think this is partially related to the pandemic, where in many cases, there was a pull forward of demands. And there was just a behavior on the part, I think, of large companies where it was like, we are not really putting a lot of scrutiny on what we're spending on software because we just want to make sure that everyone who's working from home has everything that they can possibly need, right? Like you just built a new house and you're getting all the nicest things for it, right? And you're not really paying attention to all the costs associated with it. And then as interest rates went up significantly and people became a lot more cost conscious and started tightening their belts, it turns out a lot of that pull forward of demand resulted in sort of slower budget spending behaviors on the part of companies. And the big impact there is that it can slow the growth rate down. And so that sort of double whammy of growth rates coming down and multiples declining is a big part of what happens from 21 to 22 to where we are today. And so that's a little hard to predict with a lot of precision going forward, but a big, I think, wildcard going forward if we're at the lower part of the multiple decline is what will happen to growth rates. Will they kind of pick up to where they were or get somewhere closer to that or will they stay relatively subdued? Because that will have a big impact on the pace of valuation increases going forward. Teddy? Yeah, I agree with that. It's not as if in 2021, the venture and growth investing community just decided that we will invest the same kind of assets we were doing a few years before at double and triple the multiples. It's more the fact that those assets that a lot of people saw in 2021, they looked like they were growing incredibly fast and they were kind of very unique. And of course, it turned out a lot of that was kind of demand pull forward or something that was not sustainable, but a lot of very smart people kind of underwrote that growth just to continue year in, year out. And that's kind of how we got the 2021 situation. It's also good to recognize that software sector, it's not totally uniform. There's different subcategories of software that have different kind of shapes of demand curves and over this period. Of course, the general trend is very clear. There was this big uptick and now it's like slowed down or gone back to historical kind of averages. But some sectors like cybersecurity is still very strong. We've seen strong demand throughout this period. Just to pick another one like sales and marketing software, which was selling a lot to tech companies has been much, much harder in this current environment where sales heads are being caught and budgets are under pressure. And some other categories of software are kind of in the middle. So it's also kind of important to realize that it's not like there's one market called software. There's a lot of different product categories. There's a lot of recurring customer segments. Selling to like venture backed tech companies right now is not a tremendously attractive customer segment because they are under pressure for sharing profitability and so forth. But selling to main street businesses and let's say healthcare or financial services in the US is pretty stable. Yeah, no, that's right. Some of the folks, including myself, I think that the opportunity, the gen AI opportunity that lies in front of us has the potential for being one of the largest opportunities that I've seen in my career. And I worked through the internet boom, so as to speak, and the cloud and the SaaS businesses, et cetera. And what's your take on generative AI and its place in the future? And how do you lens it specifically from a investor perspective, Teddy? I mean, I agree with you. It's clearly going to transform so many industries and so many business processes. There's this hype cycle that sometimes we as investors and founders, we tend to sometimes overly hype some things in the beginning of a trend, and then you find this plateau of productivity. I think we're going to going through that now, but it's clear that the promise is very large. The way we look at it, and Lonnie can talk a lot more about this, is that the fiction shovels of this trend is very, very interesting. You look at all the kind of new, not just the foundational models, but all the workflow and the data stack and all these things. There's just a same way with mobile and cloud. There's always a whole new trend of opportunities that's usually captured by new companies instead of incumbents that comes with this change in computing platforms. And I think that same thing's happening in AI. It doesn't necessarily mean that there's going to be 30 abstraction frameworks for LLMs or something like that, but it's clear that there's going to be a lot of new, very valuable companies built over the next five years in AI tool chain. And Lonnie, you've got a very interesting framework, which every time I want to sound intelligent, I copy and, of course, call it mine. Why don't you share your insights into what you think the future of Gen AI is likely to be? Yeah. So, I mean, we've been thinking about this a little bit as, right, so you had the, I'll call them the AI prediction system, AI recommendation systems, classification systems, the sort of post AlexNet, so after 2012, but before the foundation models came on the scene. And those systems of prediction and classification are also improving at, in some cases, a fairly rapid pace. We've been seeing some really interesting things going on in healthcare with portfolio companies like Overdead and Inventive Health and others. So I'll put those aside for now. So there are these new systems that are, in a sense, still doing prediction, right? They're predicting the next token in a body of text, or they're predicting how to modify an image in order to create something more of what you asked for. But qualitatively, it's different. They're really more systems of creation. So they can do things like write code or generate images or videos or draft blog posts. And what they're doing, effectively, sort of oversimplify it a little bit, is they're driving down the costs and price of a huge number of additional tasks, sort of like what we saw with electricity and cars and computers and even the prior generation of AI, the internet, mobile phones. It can commoditize certain business models. It can make certain human tasks less valuable. It can also make things much more valuable. It can create new business models. It can unlock talent in new places. And the stakes are very high, right? If you just imagine something like a 5% improvement in EBITDA and the S&P 500, and you put a 12x EBITDA multiple on it or something like that, you're talking about trillions of dollars in value. And it's just been very rapid, the pace at which it's starting to impact all sorts of industries that have historically been somewhat resistant to extreme impacts from software changes, right? Education and law and healthcare and media and content creation, software development, and things like that. And so the point that Teddy was hinting at around the balance of power between startups and incumbents, it actually feels pretty different from a lot of shifts in that you look at what's been happening with some of the large players who have released foundation model capabilities already this calendar year. And we've seen it with our own portfolio as well, where we have dozens and dozens of companies that started working on their first hackathon or brainstorming session, single store included in this in a way, even though it's not at the application layer. And they've already released something to production this year, right? So it's like they started working on something and it's already out, and it's only been not even a full calendar year. And often the advantage they have is multifaceted, right? So they have distribution, right? They can, you know, Adobe or Microsoft can light up a capability in PowerPoint or Photoshop to millions of users worldwide. They also have the ability to leverage data assets that they have access to, right? Adobe trained Firefly on their Getty Images competitor, where they own the copyrights of the images. And so they can identify the customers against copyright infringement, things like that. But they also have their products and people use these products, they have enormous market share. You know, when I was an intern at Microsoft working on PowerPoint, I think the market share, you know, within a couple of years got up to above 95% or something like that. And there's a lot of inertia around people wanting to move off of tools in order to use generative AI capabilities when those tools also have them, right? So if you're trying to disrupt those products, and your only angle is that you're using generative AI, and those tools also release generative AI, but the gen AI knows how to use their products really well, it's going to be really tough. So what we've been paying a lot of attention to is, all right, so what are the kinds of moats that people are building up around this technology, right? There's the obvious ones around having an existing application and distribution and data assets. Trust seems to be really important. And that also tilts towards large incumbents to some degree, that people are very anxious about these technologies for many good reasons. Brands that are able to earn trust around privacy and safety will have a pretty big advantage. This is hard to build, but easy to use trust. Building out efficient infrastructure for what people call inference, so running the systems, that seems to be a much more significant source of advantage than I would have expected, and more durable too, than I would have predicted in the sense of like, if you can have a more efficient infrastructure, if you can run your system at 90% lower cost, and you grow faster than everyone else, that advantage tends to increase over time. So you start to outpace the competitors. Having the right kind of talent, being able to build beautiful products that people love. There's some emerging moats around usage demand side economies of scale. So as more people use the product, you get feedback data, which you can use to improve everything from prompt engineering to fine tuning data, reinforcement learning, feedback, things like that. And then investors with the ability and permission to invest over long-term time horizons, where you don't have to return EBITDA or gap EPS to your investors within the same calendar year, where you have the ability to plant some EBITDA in the ground or EPS or profits in the ground to blossom into a profit flower or profit tree in a few years. So we've been thinking about, okay, that's all interesting. Does that just make this hard to invest in? And the answer is yes, but there's also some really interesting areas, some threads to pull on. So one is, a lot of these tools are what Kevin Scott for Microsoft has been calling, making hard things easier, right? But there's this class of thing, which is about making something that wasn't possible, possible. And those can represent interesting startup opportunities because by definition, there may not be an incumbent because it was impossible. And because it's still, it's possible now, but it may still be hard. So you can enjoy a technical moat for a couple of years and get to enough scale that you can eventually transition to some more scale-oriented moat, like NetworkFX or DataFX. Our portfolio company, ProFluent, which has chat GPT-like technology to create protein, do protein design, is essentially a good example there. Infrastructure software, I'm on the board of a company called RunAI that, sort of like VMware for graphics processing units, it lets you drop your GPU utilization for a really large cluster. If you're running on a big NVIDIA cluster, you can run it much more efficiently. You can let teams work together in a way that's much more efficiently. You can drop your utilization by up to 80%. Those products, when highly differentiated and start to build up platform effects and get wired in from an ecosystem perspective, those can start to build up talent network effects where people learn the product and take it with them from job to job. Those can be really interesting businesses that have massive adjustable markets. There's some interesting venture buyout opportunities, I think, as well, where you have a business that's perfectly good on its own, but they are the incumbent in their space, sort of like many of our existing portfolio companies, where they, with a little bit of help, we're building out a lot of infrastructure at the fund level around using this kind of technology internally for marketing and sales and software development, but then also how do you incorporate it into products on a detailed brass tacks kind of way? How do you leverage single stores, vector database capability to do retrieval, augmented generation in a way that's super efficient and things like that? And if you can sort of invest in, maybe even do a venture buyout of that kind of company and then upgrade them to this new product, sometimes the Gen AI product, when combined with the underlying product and install base and data asset, can be even bigger. It can be a feature but even bigger than the underlying product. And you see a little bit of this with GitHub Copilot versus GitHub. You imagine the eventual state of GitHub Copilot where it's writing an enormous portion of your code, even though it's a feature on top of the IDE, the software development environment, in some ways it's a much bigger thing than the underlying product that it's built on. And you can potentially charge multiples of the underlying product for it. And then I would just wrap up with, there are a number of challenges here, right? So things like hallucination and alignment and a lot of legal questions, a lot of issues around explainability and privacy and training and inference costs being very high and things like that. And these challenges represent startup opportunities, right? And so to the extent that people can solve them in a way that's differentiated, that's an interesting additional investment angle. And there's just a lot of massive amounts of unknowns. I'd say a really big one is just how quickly will we reach a state of diminishing returns and capability to these language models, will it be really soon or will it take a really long time? People will act like they know the answer to that, but they don't. There's a lot of those kinds of issues, large models versus small open source models, which ones will win out in the longterm? How should AI be regulated? Should you do fine tuning versus vector search and prompt enhancements? Will startups or incumbents capture more value in this area? Will it increase profitability and growth and drive down prices? Things like that. And so there's just a lot, I think you can, there's a lot you can do from an investment perspective by just watching the way those things are playing out and keeping an open mind about them as opposed to being stubborn and jumping to conclusions before we know all the answers. But if I think about this, a lot of the new application startups we see are AI driven. I think what all the founders underestimate is how sticky workflow is in B2B scenarios and integration of workflow. Just because your thing is easier to use and faster because of AI, doesn't mean that it's going to be successful. Take an extreme example. Think about when you go rent a car and they just type away for 15 minutes without saying anything. They're using a terminal-based software that has changed its UI for 25 years. This amount of inertia in enterprises in the systems they deploy in the application area is just enormous and something would have to be so much better for many enterprises to go through the pain of changing an UI or changing a system because of all dependencies. And that's something like some founders are still missing when they are looking at how to apply AI to disrupt X category. Yeah, that actually brings me to the last question of this fascinating chat. So Teddy, what do you think are your predictions for 2024 in the tech world? Top three predictions. I mean, this is like an AI next token kind of prediction question, but I think what we'll get in 2024 is sort of more of the same. I'm kind of more on the soft landing camp. I think we'll just have a gradual recovery in tech markets. I don't think we'll see anything dramatic to the upside, but I'm also expecting and hopeful that we won't see anything big to the downside. So what it means is that we'll see slowly growing software spend. Things will continue to be somewhat hard, but kind of constructive. I think the kind of financing and investing is going to be very fundamentals driven. So companies will have to show good balance of growth and profitability, but investments will happen and runs will happen. So I think it's more of the same in a way that is slightly geared to the positive. That's my expectation. Lonnie? Yeah, I agree with those. Just on the AI thread, I would say I think the major focus, at least at the beginning part of the year, is going to be around incorporating a lot of these technologies into existing software products and packaged software. So there's a huge number of large software vendors who haven't released their Gen AI offering yet. But when they do, some of those things will be really interesting. People will maybe stop building their jury-rigged solution to that issue and start using it and realize that the combination of existing powerful software, data assets, and these capabilities will be quite impressive. Most people still haven't even gotten access to the Microsoft Office Copilot capabilities yet. They still write their own PowerPoint presentations like fools. So that I think will be a big activity next year. I think we may start to see some early traction around things like new form factors for AI devices. I don't know what those will be. Are they glasses or earbuds or something you wear on your lapel or some thing that gets attached to your phone or something we haven't thought of yet. But I think you start to see a little bit of the ghost in that machine if you play around with the chat GPT voice conversation where you'll talk to it and it talks back to you and you talk to it and it talks back to you and you brainstorm with voice. Even though in some ways it's just a user interface tweak, it feels really different when you're using it. And related to that, I think around the devices, I think there's a lot of new applications that are coming online with the more native multimodal functionality. The ability of the systems to understand images and then incorporate that into all of the knowledge that it has already and produce responses that include both text and also images and eventually video. I was having an issue with the electric blinds in my apartment and took a picture of the device that stuck on the wall and brainstormed with chat GPT to solve the problem. And the first couple of approaches didn't work and eventually it wrestled it to the ground and it was an amazing way to solve a problem where you can just sort of talk to it. And then I think we're going to see some of the industries that have previously been very resistant to softwareification. Software, I think over the last fairly long period of time, has been an unusual part of the economy because if you compare it to other segments that people have traditionally thought of as large, like banking or telecommunications, which have tended to have just a few really big companies, the software market creates more billion-dollar winners than pretty much any other industry. But one of the ways that it does that is by sort of taking over parts of other industries like Netflix and entertainment or Amazon and retail. But they're really companies that have software DNA at their core. But there have been some industries that have been pretty resistant, like healthcare and education and law and things like that. And I think these new capabilities open up those frontiers a bit more. And really what you need is reliability to improve. And there's a lot of legal issues that people are struggling with. But also just at some point, you hit this tipping point where the system becomes better than the people. When you get to better than human performance at some task, it's no longer a minor efficiency improvement anymore. At that point, it's like a phase change, like ice turning to water. And everything changes. It doesn't make sense to do it the way you were doing it before. And people start to reorient around it. We may start to see some example use cases there in 2024. And then it wouldn't surprise me if we see some motion, even though this is much harder, around advanced robotics. You see early versions of this with Boston Dynamics and what Tesla's doing with their humanoid robot and some things like that. But it's still pretty rudimentary compared to the amount of progress that's being made on the language front. But we may see some more progress there. And then some really interesting things around the combination of AI technologies and things like drug discovery and bioinformatics and pharma, where it seems to be the combination of these models and really sophisticated search and more AlphaGo or AlphaZero style reinforcement learning can start to basically leverage the capabilities to do brainstorming and then searches and attempts at creating answers. And it can actually mimic some aspects of the scientific discovery process and maybe come up with truly novel things. And bioinformatics and drug discovery seem like one of those frontiers. That's fascinating. Thank you, gentlemen. Teddy, Lonnie, thanks for your continued support. Thanks for your time today. I'm sure the listeners would find it fascinating and wish you very happy holidays and a very happy new year. And hopefully 2024 is slightly better than 2023. It's, you know, I always said that it's been at the bottom for about six to nine months of the year. And hopefully this bottom, you know, there is an upward trend next year. And cheers to that. So thank you all very much. Indeed. Thanks for having us. Thanks very much.