The Infra Pod - Infra Reflection on 2023 with Tim and Ian!

Episode Date: January 8, 2024

Tim and Ian sat down to talk about what we learned in 2023 and our reflection on Infrastructure companies and projects as a whole! Come sit down and chat with us :) Feel free to post yours on X and t...ag us @tnachen and @ianlivingstone!

Transcript
Discussion (0)
Starting point is 00:00:00 Welcome to the last episode of the year for yet another Infra Deep Dive. This will be a fun one because it's just me and Ian. So as usual, I'm Tim from Essence VC. And let's go, Ian. And I'm Ian Livingston, currently trying to help Sneak figure out how to become a platform. And I couldn't be more excited, Tim, to go through our spiciest learnings of the year and then talk about next year. You and I have had so much fun learning with everybody else
Starting point is 00:00:34 about the future of infrastructure. So I thought it would be a good idea. I messaged you a week ago and said, we should really do one of those recap episodes. And here we are. So do you want to kick this off and talk about our top three learnings? Yeah, we should do it. I just want to mention, I don't think I've ever done this before, so it'll be fun. It'll be fun. It might be a total hit the wall and crash and burn, or it could be amazing.
Starting point is 00:00:59 We'll all find out and learn together. We'll figure it out. We'll find out. Exactly. All right. I must figure it out. We'll find out. Exactly. Yeah. All right. I must say it. I think the biggest takeaway from 2023 for everyone, hands down, is that cost matters again. And I think everyone started off in quarter one of 2023
Starting point is 00:01:18 with basically marching orders on how can we rationalize how much we spend on making things work, right? How can we rationalize how much we spend on making things work? How can we rationalize our cloud spend? How can we figure out what people we need in our engine organization? What we don't? What can we cut? We're back at a point in society, which I think is a good place for us all to be, where capital has a cost. It's not infinite and free. Now when we make a decision, we have to think about, how much is it going to cost
Starting point is 00:01:45 us to run? And as a result of that, is it good for the environment or not? As a result of that, is this actually an innovation or not? We spent so many years, specifically on computers, on the infrastructure side, in a world where we could legitimately get as much capital as we needed. And so the only focus was, how do we fill top of funnel? It doesn't matter how much it costs to operate, because capital was quite cheap, and now capital is a cost. And so I think my greatest takeaway is both one of the best parts, it's not just about performance, performance, performance,
Starting point is 00:02:17 but at what cost. Now this forces us to make better decisions, build better infrastructure, and have better informed decisions across the board on infrastructure we create, but also the infrastructure we use. I totally agree. And cost has always mattered, right? But it's really, like you said, some people have decided that the capital costs on actually buying infrastructure software probably doesn't really matter as much. When you say costs, there's many different kinds of costs, though, right? There's like the cost of education to get developers running something. There's a cost of maintaining of human costs. There's a cost of speed. Back in the Netflix, I was talking to
Starting point is 00:02:55 Capital One, right? They moved to the cloud so fast, fully betting on cloud. It's like, okay, are they dumb? Are they smart? There is a cost to it. Actually, I feel like the net cost of running that infra from a pure compute cost, probably higher, but the speed of development costs probably lower. Fundamentally, infrastructure cost is really thinking about the cost of the money and time and energy required to really decide what matters here more on some factors, right? If you're a high grow business and margins are like, you know, Netflix revival probably doesn't matter as much as margins. Like 5% difference really matter. 10% difference doesn't matter. Grow at all costs to some degree, right? But it's like the squeezing margins really happen now as much. So I feel like
Starting point is 00:03:44 cost will still come back to have different trade-offs. It's just sometimes really hard to actually know what kind of trade-offs we're actually making. Yeah, and I think to your point, the number one thing we cared about prior to 2023, like in the zero interest rate environment that we had, was just velocity.
Starting point is 00:03:59 It was a complete focus on how quickly can I integrate, how quickly can I roll it out, and how quickly can I build a feature or get some improvement with the door to compete with. Because that's all that mattered because capital didn't have the same cost, at least in startup land. Now, an operational business trying to yield a dividend or where the street is marking it on cash flow, there were still those businesses in the world, but the vast majority of tech didn't live there. The vast majority of tech was just riding that growth wave. And now we're at a point where, oh, it's also about that runtime with the unique economics of delivering whatever it is that you're delivering. Whether it's receipt, it would be scanning a piece of software and telling you vulnerabilities. For Uber, it's the unit economics around how much extra compute cost is involved in routing traffic on top of the cost of the actual person, like getting the actual person to drive the car to get the
Starting point is 00:04:47 delivery. All of those things, it's unit economics that we all now really focus on because it's the thing that drives the primary margin at the end of the day for the business. And that is different than some of the decisions we make before. Now we're going to be very specific in terms of the instance size we choose, whereas before we wouldn't be as specific. So there's all these things, this whole sort of world we've gone through. Now cost matters. And I think the biggest place where we haven't yet crossed that bridge, but we'll cross that bridge, is the cost to deal with large quantities of data. For years it's been, well, just solve the problem because we need to move on with our lives.
Starting point is 00:05:20 And I'm interested to see what this new world of cost, how that opens up, and how really smart founders, infrastructure builders, take advantage of this cost-sensitive world. And I think the greatest example of this are the Warpstream folks we had on a previous podcast and how they're approaching building a new version of Kafka. And I think it's just the greatest and most interesting time to be a hyper-smart founder focused on building a next-generation piece of infrastructure and knowing that you can out-beat the competition by having a significantly better architecture that yields a way better unit economics. Yeah. A ton of thoughts here, but let's move on.
Starting point is 00:05:53 I think we have plenty to talk about. I think from my side, if you ask me at the beginning of the year that there will be a company like Fly.io or Modo that can grow and be successful, I will be like, do we need another company that does deployment? Do we need another serverless company that can run functions? Like one thing I didn't really internalize, it's funny because I worked on like Docker and
Starting point is 00:06:17 Kubernetes stuff. So I kind of like understand, but it's a good reminder at least that, I mean, they're not like IPO and huge X's already, but they are growing pretty successfully, right? One thing I learned from these is that it's hard to use past history of what products look like to try to predict. Serverless, for a while, only means a certain kind of workloads. And for a while, it means a certain kind of experience. And that was capping how we use serverless and how we use certain things. And the flyers of the world
Starting point is 00:06:49 really made deployment so much more easier. Everybody flocked to it. It's not like the alternatives really doesn't exist at all. The alternatives are always there. As developers, I really learned that if you can reduce friction 10x better in any shape or form, I think that's a product that just magically get people to use it.
Starting point is 00:07:09 Because if you go to ask developers, would you want to use a new serverless product? Or would you want to use a new deployment to run a server somewhere? Like, ah, we have that. We have a script. It's already finally figured out. It's not easy. It's hard, right? You cannot ask developers what they want from a high level.
Starting point is 00:07:27 It's really the other way around. It's like once the magical experience is there, people flock to it. And that's always amazing for me to see. You know, at least one thing I learned is that developer experience is going to be a constant changing bar. Somebody that has a good developer experience 10 years ago, five years ago, even last
Starting point is 00:07:45 year, doesn't mean the same thing can just happen again, right? Whoever is going to be the best developer experience got to reduce the friction of existing alternatives to like a pretty major factor. And if you can just do that, you'll be well. It doesn't actually need to have any crazy pain points. You will have to deal with that at some point. But I think as a starting point, that's a very good measure to see if something will be adopted quickly. I know it's probably obvious,
Starting point is 00:08:14 but as a developer over time, you're just like, oh, serverless kind of like never really took off, you know? There's so many deployment options. Do we really need another one? I guess, you know, back in the day, search engines, there's a bunch of them. Google came up and it just crushed everybody, right? You know, I think there is always an opportunity for someone to actually completely change a game.
Starting point is 00:08:34 It just doesn't, will not just look like yet another serverless or yet another deployment. So it's nuanced, but I feel like that initial thing, if you can see the difference, I think it's actually one of, can be the most exciting thing to happen in this development world. That's a great take. I often think of developer experience almost like a prosumer use case. It's like what you're actually trying to create is a feeling of speed and elegance.
Starting point is 00:08:59 And you're trying to remove paper cuts and hiccups. And the more paper cuts and hiccups and friction, and the more speed you create, and the more natural it feels, the better the DX. It's like this intangible thing where you can't describe or measure it, really. You can, there are ways to measure it. But at the end of the day, the reason I made
Starting point is 00:09:21 Buy a Little Lemon versus going off Amazon is there's a vibe and a feel to it. It's also about brand. It's about all these different things. It's a sensation that a developer has and how they identify with the product and what that means about that person, why they use it and how they identify and also how it feels when they use a product and how that enables them to be successful. And this is what so many people get wrong when they think about developer experience.
Starting point is 00:09:44 They think, okay, well, it integrates to the pull request or it's an IDE plugin. It's like, no, that's just checking the box, bare minimum enterprise sales. It's like it's the actual best developer experience leaves the developer feeling like they have superpowers in a way that you can't get if you're writing a deployment script on Kubernetes.
Starting point is 00:10:02 Yeah, and I have more to say, but I'll leave it to the hot take section. We'll keep some of that there. All right, man, it's your turn. Okay, this one, I got to be honest. For those that know, I spent a couple years trying to build a data privacy company. The whole thesis was that proprietary data
Starting point is 00:10:17 is the only thing that matters. And when the LLM revolution hit us late last year, I was like, well, the only people that will win are going to be the big five. I think with Onidote, early part of the year, we all saw a surge of AI features coming out of every one of the big tech companies. And it felt like this was over already. And my whole thesis was like, well, they have all the proprietary data. They have all the training infrastructure.
Starting point is 00:10:41 They have everything they need. And I think the thing that I didn't anticipate, and I will be completely honest with you, I didn't anticipate the fact that the way they'd fight with each other, for example, Facebook and Lama, is that they just start open sourcing the models. And as a result of that, my thought process around who will win in AI and how AI felt like it would be a net win for the dominance syntax. I don't believe it anymore. There is some truth to proprietary data. It does give you an ability to find a model for a specific use case that's better. That is only true. But the reality is that
Starting point is 00:11:15 with all of the open source tools and all of the models that are out there, this is one of the most democratized revolutions or infrastructure waves that we've ever experienced. Kubernetes democratized our ability to run service-oriented architectures without a doubt. But all that lived inside Google for 10 years with Borg, before Kubernetes came out. So I often thought the job of a great infrastructure company
Starting point is 00:11:40 was to democratize access to stuff that Google had already written. And in this case, we're seeing this stuff in real time. And the result of all of this is, there is a capacity to beat the large incumbents at their own game, not through proprietary data, but through the open source ecosystem. And I was shocked near the end of the year when I came to that realization, and I will eat my own hat and tell you, at the end of the day, I don't think this is just a wave
Starting point is 00:12:04 that the big five companies are going to ride. I think this is a wave we can all ride. And I think that's incredibly exciting. Yeah. You know, I was reminiscing a little bit with D2IQ shutting down, which is one of my earlier startups I joined, or Mesosphere. It just reminds me that the whole container wave of companies, Docker is arguably the biggest winner in the independent
Starting point is 00:12:26 startup land. Of course, everybody has an EKS, GKE, every large company, and everybody else wins by democratizing Kubernetes. In every startup, really the best we've seen is just a couple hundred million exits to some acquire, and that's it.
Starting point is 00:12:42 Or Mesosphere that was so hot for a couple years, and that's it. And then, you know, or Mesosphere that was so hot for a couple of years and with almost nothing. I'm sure the LLM space is going to be different, but it's always a good reminder that hot and valuable is not correlated all the time. And value creation and value aggregation still really, really matters.
Starting point is 00:13:01 I do think that there are opportunities for all of these startups to have also an exit that is actually quite meaningful and they never took it, too. So for founders out there, sometimes you've got to take the offer. And your VCs are not going to help you make that decision, unfortunately. You've really got to think through the defensibility of your business. I think Docker's a great story. It's not Docker Swarm or any Docker's registry that has turned Docker into a viable business.
Starting point is 00:13:29 It's the fact that they have a desktop app that's pretty good and they're able to charge businesses money for using it. They had crazy distribution of that desktop app before they turned their money spigot on. And now because of that distribution channel, they can shove value-add features for Docker using use of containers at customers. But it's incredible to say that you're right.
Starting point is 00:13:49 You're always shocked by who actually wins the ecosystem. If you were to say the Docker of 2012 or 2014, where the Docker is today, the cap table is definitely not the same at all. It's a completely different company. And that has yielded some pretty exciting stuff for them. And I think we'll see the same thing in the It's a completely different company and that has yielded some pretty exciting stuff for them and I think we'll see the same thing in the LLM world. For sure. Maybe that's already my second point.
Starting point is 00:14:13 It'll be very difficult for any LLM infra company to operate. It's not impossible. I think there definitely are cases you can win, but it would not look like a horizontal LLM infra. You can just say everybody use and everybody loves it. I think lane chain is like the closest we've seen so far. And we'll see the jury's still out.
Starting point is 00:14:29 What's going to happen to this? Maybe this is already a hot take, but, you know, LM infra would definitely require a lot more thoughtful thinking to remain your competitive edge. And the hardest part right now is everybody's going to eat everybody's lunch. Because no matter where you come in, it doesn't matter your evaluation, your monitoring, your prompts engineered, your this and that,
Starting point is 00:14:52 just one hop away, you're going to get overlap with someone. It's like the modern data stack times five. In the modern data stack, we see like five, 10 boxes. They're all just lurking around. And MLOps was similar. There's like 10 boxes,
Starting point is 00:15:06 everybody just kind of lurking around. MLOps was similar. There's like 10 boxes everybody's just kind of lurking around and in the end like nobody won. No. DataRobot maybe-ish? Like still still nothing. I don't know. Still nothing won.
Starting point is 00:15:16 So it is quite sad to see. LMInfo has more differentiation to some degree but it's not going to be easier to operate. So it is quite shocking and still really hard for anyone to really find true differentiation. And, you know, if you're a founder, you're really also hoping for the best. There's too hard to predict what's going to happen in the future anyways, right?
Starting point is 00:15:37 So a lot of people come in with the hype, you know, just like having people come into the crypto space with the hype. And we'll see, sir. Something will happen here. Well, at the end of the day, I think the one point I'll make on this is you've got to decide what business you're in. Are you in the business of selling network, compute, and storage? Or are you on the business of selling collaboration software?
Starting point is 00:15:58 And you're either one or the other. And there's a lot of value-added that can happen on the collaboration side. But if you're selling raw compute, raw storage, and raw network, which is a lot of what a lot of the container companies were trying to sell or the LLM infrastructure companies were going to sell, the truth is that Amazon, Google, Microsoft, Cloudflare, these are at-scale operating businesses with incredibly good margins that it's hard to compete. You're going to lose to the biggest fish in the room unless you can get the capital to drive those costs down.
Starting point is 00:16:26 And I think that's the most important takeaway I have for any founder is decide what game you're playing. And if you're in a capital-intensive game, know that there's a ticking clock before the big fish catch up to you. Yeah, it's so tough though. Anyway, I think that's a big topic we can discuss. What's your number three, man?
Starting point is 00:16:44 My number three, and I will completely throw this out to the wind, but React server components being as important and as revolutionary as they are, and the fact that the heyday of PHP strikes again as a result, I was surprised. It's not surprising in retrospect. I think we're always going to get back to here. It's always clear that rendering stuff in the server
Starting point is 00:17:03 and sending it over to the client prior to, so it speeds up what the work needs to be on the client to get the full page load makes complete sense. But it's just an interesting case study in the continuous history of software infrastructure where we go through these cycles of thinking we're building something new, it being single-page apps,
Starting point is 00:17:20 single-page apps not having the performance characteristics we want. So then we go back to trying to figure out how do we have our cake and eat it too and ultimately end up with things like React-Service-Depont. So I think React-Service-Deponts are great, but at the same time, I will be honest, I was surprised that I didn't see this coming out, but I think it's one of the most important things that I saw in early 2023. Yeah, yeah. I don't know too much about React,
Starting point is 00:17:42 so it's hard to comment on that one. One thing I definitely feel like will be interesting to watch Yeah, yeah. I don't know too much about React, so it's hard to comment on that one. One thing I definitely feel like will be interesting to watch is the trend of this front end becoming full stack. And we did a very first episode with a little chatter, hoping to get a little bit more insight. I think it's too early to tell what the end state of this will be.
Starting point is 00:17:59 But I do think looking at what people are doing right now in companies, this is maybe like a hot take with what I hope to see as well. Like, is there a fundamentally better way to just write apps? You know, like this whole way of doing React plus this tool plus that tool. Because if you look at what fundamentally what the new startups and dev tools look like, there's going to be the pydantic, rough type of people out there trying to consolidate tools just on the front end, hoping to kind of creep into the app side. We're also seeing, like, I was talking to Gatsby founder
Starting point is 00:18:33 on our open source podcast, and seeing how that journey went out and competed with Roussel and Netlify. I think there's like waves of frameworks that are trying to compete with one each other to just eat spaces in their tool chain and hoping to accrue value somewhere. And I think it's a valuable thing to consolidate.
Starting point is 00:18:55 I just don't know how it's going to happen. One is like, I don't think the consolidation shouldn't just be doing something that does, because, you know, rough and pydantic to some degree is like doing what's happening already but faster. And there's going to be like more revolutionary approaches like the local first type or changing how we react
Starting point is 00:19:12 and how we state it's being synced and stuff. Is that only meant for certain kind of applications or can we all change how we write apps? Is there going to be like a big bazooka moment almost, you know, like every single page app will be done differently. I feel like we're heading towards a direction that people are just craving for a big
Starting point is 00:19:30 change. I want to be see a fundamental different approach when it comes to building apps from day one. Like not just like come to learn React, know how to write the Redux stays, architect this, architect that, think of API, think of this, think of that. I think we're all doing this because of familiarity and some level of flexibility in the end. It's naturally because these are the best tools ever can be invented. I'm hoping next generation, maybe hopefully next year or two, full web apps will be done just fundamentally differently. The starting point will not be like a blank JavaScript page with a React init, right? You know, and like that starting point always not be like a blank javascript page with a react in its rights you know and and like that starting point always looked the same can we change how that looks like
Starting point is 00:20:11 anyway i'm seeing a lot of people trying to change i don't know how it's going to change fundamentally but i'll be excited if something could actually change it a lot where complexity reduces so low yeah when i almost often think of this problem, I think at some point there's so many people that understand the current process and madness of this. It's like, how hard is it to change inertia? So I'd love to see a new revolution in the way that we build front-end applications. We kind of have gotten there.
Starting point is 00:20:37 React really was revolutionary at the time. And we've seen the fact that we can now cobble together components and then manage the interaction between those components in that state, what that enables you to do. I would love to see a change, just for the interest's sake. But I also wonder, we may be close to solving the actual problem
Starting point is 00:20:56 of all the problems. Do you think Wasm is part of it? Oh, for sure. Wasm helps take complex pieces of code and move them around. And that's the best part of Wasm, is interoperability. So I think Wasm will help us build better web apps. And I think Wasm will help you build better apps in general that exist across multiple different services
Starting point is 00:21:14 and are thus more consistent. So Wasm is a part of that. But I don't think it finally will change. In terms of components, are one of the best ways to think about cobbling together a UI and the form factor that is HTML. So I will be excited, but I also think we need a change in UI. Like the way that we actually interact to really get the change
Starting point is 00:21:34 in the way that we build stuff. Yeah, I feel like Wasm is really one of the most exciting, but to me, it's also the most confusing part of the stack. People are hoping this thing will revolutionize and change everything, but it's so low level to a point that it seems like it's a great starting point almost. Like sure, I can compile my code to Wasm. I can take a database, run it in my browser. That's like a starting point.
Starting point is 00:21:59 You know, it's not going to truly just change everything. There's so much friction right now. I think there's a lot of room to move. And like I said earlier, whoever reduces existing friction down 10x, you're going to see a huge amount of people flock to it. And I think there's going to be somebody able to figure out the moving pieces here,
Starting point is 00:22:17 which will be quite interesting to see. Well, I don't know we have time for three hot takes. Do you want to do your hot take? At least one hot take for next year? Oh, okay. Spicy features. All right, here's my hot take. I've got so many more.
Starting point is 00:22:35 I think the reality is that the role of LOM to production is going to be completely hindered by the fact that we don't have a good way to test and upgrade the model we're using. So as the model changes underneath you, it's hard to engineer the prompts and control the service area. I think this is a feature for consumer apps because it allows creativity and generation, and I think it's a complete liability for business apps.
Starting point is 00:22:58 And you can kind of already see that happening if you're following along on Twitter. There's a lot of memes going around where people are using ChatGPT-based chatbots to ask random questions via the help pop-up, the contact support pop-up on random websites or probably ChatGPT. So I do believe that we're going to reach a point where we haven't yet actually figured out how to build reliable applications with LLMs. And that will hinder its adoption, but also hinder the complexity of the application we can build with them until we figure out a better way to basically test, upgrade, and roll out changes
Starting point is 00:23:32 to an L1-based app and know that it will work in the way that we want it to work with confidence. That's my spicy take is, hey, this is pretty cool, but also, while it's a feature for consumers, it's definitely not a feature in the business case. It's actually actually in many ways probably going to end up being a liability. And so maybe we end up back at decision trees
Starting point is 00:23:50 for some of these use cases. XG Boost. XG Boost, yeah. You know, I think my related hot take, I think given this AI wave is so huge, we're talking about element infra already, but I think there's also a set of Infra products that are like AI native. AI, SRE, AI, I don't know, monitoring, AI. At the end of the day, I fundamentally feel like companies that will succeed for the next
Starting point is 00:24:16 generation Infra products, it should be as exciting to use without AI. Like it can't just be like I'm an AI native and I have AI anywhere in my pages and I can do so many AI stuff. I don't think Infra is the place you want to put AI anywhere easily at all. But we see so many AI everywhere products now. You know, like I tell you root causes, AI tell you this, AI tell you how to say hi to your server.
Starting point is 00:24:42 I don't know. There's like this weird trying to like marry LLM into every part of the stack. And we see AI already in networking and stuff. Like this is very carefully tuned, very carefully placed. Today, most people are just like putting it anywhere. Because the demos are easy. You know, like, well, look at that. It found that my server crashed because there's a memory exception.
Starting point is 00:25:04 Of course, LLM can figure that out. I think just most people kind of forgot or never maybe even lived enough in the hot seats of being under pressure and fixing a server under 3am with all the unreliable components telling you this random stuff. Anyway, my hot take is I think any infra product for it to be the next big thing should be as exciting without AI. And then you can add AI anywhere, sure. But don't say I'm an AI native where without AI, you're almost nothing. That's why I'm actually, I'm personally very excited about products today, just like plain old infra products that can add AI, but that's not the only thing you're selling. I couldn't agree more.
Starting point is 00:25:48 Determinism is the delightful feature. One of the best parts of the microprocessor is that you give it two numbers to help to add it, and it'll give you the same answer over and over and over again. It's incredible that that's the way that works. In truth, most of the world, we want to work that way.
Starting point is 00:26:03 If we could, we'd want natural language to work that way as well. It's just a really hard problem to solve. And so LLMs have opened up the ability for us to take massive data sets and encode them lossily. But at the same time, because of that process, and then you add any other extra data to it, you change the encoding. Therefore, you change the results. You change the outputs. And what X plus Y equals Z is not necessarily true from model version one compared to model version two. You may get different answers.
Starting point is 00:26:33 And that is a huge problem when you're trying to run a business, especially a business. To go back to my first take, it has to produce margins and has liability and has customers needs to keep happy and run at scale. And I actually think while this AI cycle is going to create a buying cycle because we will build better apps that feel more human,
Starting point is 00:26:57 easier to use in many different ways, the real winners, the ones who are going to capture most of the money aren't going to be AI infrastructure. Yeah, it's a new category. But Amazon and GCP will release their own vector stores and they will make langchain-type workflows easier to do as well. The real winners are going to be the same old people
Starting point is 00:27:15 building better infra with defensible businesses. And they will ride the wave because they'll build infra that looks like it fits natively into the AI stack, but is just plain old boring infra that is just better than what is on offer. And better by a good margin. Yeah, yeah. It's tough.
Starting point is 00:27:32 We'll see what happens. I think a lot of infra companies, definitely it's not going to be all, it's going to turn out to be Microsoft for the world next year. But you know, it won't be tiny. I think that's the good news. It won't be just like everybody just crashed and burned.
Starting point is 00:27:43 There will be a few winners, just like the container space, we had a few winners, just like the container space. We had a few winners, you know? So the winner is not going to be completely zero. But yeah, you know, expecting to be the next Cloudflare, you know, IPO company.
Starting point is 00:27:54 We'll see. Something dramatically has to happen. Vector database has to fundamentally be different. Approach has to be fundamentally different. And that's the bet a lot of people are making. Yeah.
Starting point is 00:28:03 It's interesting, sir. Well, I guess this is it for our year end 2023. Yet another infrared deep dive. I guess quick plug, we are going to be having lots of fun next year as well. You know, we're going to have more guests. If you've got any suggestions, any feedback for us on what you like to hear,
Starting point is 00:28:23 what you want to do, we're really open and excited to hear what you like to hear, what you want to do. We're really open and excited to hear what you like to hear from us. I couldn't be more excited for another year of chatting to smart folks with you, Tim, and learning along the way. And we're always working to evolve.
Starting point is 00:28:38 This was a fly-by-night operation and continues to be. But we're enjoying it. We're not going anywhere, and we're continuing to get better at the same time. That's're enjoying it. We're not going anywhere. And we're continuing to get better at the same time. That's right, sir. That's right. All right.
Starting point is 00:28:51 Ciao, guys. Thanks for tuning in.

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.