Village Global Podcast - Why Nobody Needs Metrics or Traces Anymore _ Sherwood Callaway _Sazabi_

0:00Sazabi's mission is to be the default observability solution for companies of all sizes during the AI era. And our vision is a world where software heals and improves itself without human intervention or direction. Self-healing can probably be defined as a software that changes itself in response to some stimulus in production, probably like an outage or a bug or an error. I'm a big believer in creating...

0:26here we are at Village Global, sort of creating a village. I want as many allies as possible. I want people to be bought into my success, into the company's success, and to take them along for the journey. Welcome to the Village Global podcast. I'm Lindsay Pettingill, investment partner at Village Global. My guest today is Sherwood Callaway, founder and CEO of Sizzabi.

0:53Sherwood is a repeat YC founder and early engineer at Brex, where he built the infrastructure and observability systems that kept one of the fastest growing fintechs in the world running. He knows what it looks like when complex systems break at scale. Today, we're going to talk about what AI coding tools are actually doing to the software development lifecycle.

1:15Not the hype version, but the messy reality of what happens when code gets generated faster than anyone can track it in production. We'll get into Sherwood's contrarian take on observability and what self-healing software actually means in practice. This is a conversation for people who build software or invest in the people who do. Sherwood, welcome to the show. Hey, look at that. Here we go. I'm so excited.

1:40Likewise. Yeah. We're gonna, you know, go over hills and valleys, everything in between. Would love to start off with you just talking about, you know, who you are, where you're from, like, you know, we spend so much time talking about where we are today, but I think so much of that is informed by, you know, where we've come from. So, you know, you can take this in so many directions, but I'm really curious to hear, you know, 10-year-old Sherwood, what would he say about where we are today?

2:10Gosh, yeah, I'm not sure ten ten year old sure was pretty judgmental. I think I think you had a high bar for what like a well-lived life looks like I think What I was probably most interested in at ten was guitar and rock music or computers and online games and I was decidedly not interested in any of the you know the things that my parents did my dad was a doctor I Was a little rebellious and I think you know what did that look like that just meant it meant frequently being in the principal's office Usually for innocuous things It's you know, man getting getting all my parents nerves quite a bit. I think that you know, I grew up in North Carolina and You know was basically kind of a punk and and like preppy the upper middle class North Carolina and

3:05really did not resonate with the culture there. And so I think for a long time I felt like an outsider or like I wasn't where I belonged. I remember being ambitious for a long time, wanting to do something great or to have an impact, but at some point that was like starting a great band or being a musician. Who were the influences? What kind of music were you listening to? I've always loved like hard rock, metal, punk rock, post-hardcore. I mean, just this morning I was listening to early Fall Out Boy, early Green Day. Shout out to Green Day, Bay Area Band. I think in some sense my musical taste has regressed. So I've gotten older. We see comfort when we're founding something. Exactly, yeah. And I don't have a lot of time to discover right now. So I reach for old and familiar music.

4:00I mean, when you start your day with a hard rock Green Day song every day, I think it probably says something about you. Yeah, you're showing up. Yeah. Yeah. So you also played sports, right? Yeah. So I guess I'm curious about, did that counterbalance the feelings of not fitting in? Like, was it orthogonal to all of that?

4:24Because sports are such a part of Southern culture, right? Yeah, I guess so. Non-modal sports. I mean, I played ice hockey, which maybe is actually archetypally not a fit for growing up in North Carolina. I always grew up playing contact sports. My dad's from Tennessee and my mom's from North Carolina, so they aren't exactly the type to be concerned about. 10 questions? That's a big question.

4:49I think so the concussion revelations happened as I was a teenager I had like peers who were who had concussions and actually had some some serious repercussions and then also I think that it gained a lot of press broadly and there was general consciousness around You know the risks of contact sports that happened as I was Coming of age Yeah, but you know for whatever reason I grew up playing a lot of contact sports and I grew up playing a lot of team sports And I think my advantage or my strength was always I was strong and fast Uh, it was never technical skill. And I actually think that that is a sort of an analogy for, you know, a lot of my career and for what I'm building today. It's usually, uh, uh, there's a lack of sophistication associated with that. And, um, I, I look for, I look for simple.

5:38Effective solutions like the higher-order bit that often is more important than some cleverness That kind of beats around the point. You know we talked about the bitter lesson. Yeah, yeah now I feel like this there's something there's some analogy or metaphor there, uh-huh So you're bringing the footballers or, yes. It's a rough and tumble start-up AI mentality. But we see this in your manifesto, which you can talk a little bit about. From an investor standpoint, we usually see a deck. We might see a memo. But yours was the first memo I ever got that was too big to attach in my Gmail. I had to do a special...

6:20Link which I mentioned this because you know to my mind as your investor as an investor it it shows to me the the depths You know your obsession with this problem, right? Like you we usually get a three four maybe five page memo I mean this was like what 30 40 pages or something that that yeah, I mean probably should have been five pages But you know there's a lot of you know you start pulling on the thread and you think about of course you're thinking about the the core theses, the problem, the solution, what the product should look like, who it's for, how do you go to market, how you should price it, how you should position it relative to existing solutions or new up-and-coming competing solutions. And once you start asking questions, the list of questions is endless. And for me, that memo was as much for me as it was for Village and for other seed investors.

7:13Was it like a Jack Kerouac inspired and Fetamine driven like one session just? No, I chipped away at it. It was more like sculpture. You know, we have like a block and like every day I come back and be like, look at it from a different angle and be like, well, what about this? And that would lead to me adding another section. And I think, you know, the essential ideas can probably fit on a couple of pages. But, you know, I want to really have All of my eyes dotted, my teeth crossed. I want to know what I'm getting into. I think a lot of these things I also felt very intuitively and already knew, but hadn't articulated before. So it was useful for me to get them down on paper. Did you do anything like this the first time you founded a company? Because you're a second-time founder operating in what I'd call your zone of genius versus the first time. You can share a little bit about that if you'd like, but did that part feel distinctly different?

8:08I guess for context, my first company was a health care voice AI company. We started out as revenue cycle management and then built a call center in the Philippines and LLMs came around and we realized these are very useful for quality assurance and for data entry and eventually for automating the full call using a voice agent. This was a really weird thing for me to start building, being a infrastructure engineer at Brex.

8:32But I was also pretty young and I wanted to do a startup for basically my entire adult life. I wanted to do a YC venture backed tech startup because when I first learned about startups in tech, all of my peers wanted to, you know, they're reading hacker news and they wanted to work at YC companies. So it was my dream to do my own YC startup. And I was looking for something to do during COVID.

8:56And I was living with my best friend, who's also an early Brex engineer and has been a founder in the past. So you could say that we were very primed to start a company. I started this particular company because I felt like my dad being an orthopedic surgeon was an unfair advantage. And I was interested in fintech, and I thought there was an opportunity in vertical fintech. And healthcare was the vertical that made sense to me. But it was kind of an MBA case study approach to starting a business.

9:25Yeah, like Mackenzie says, this is a big problem. And the Bureau of Labor Statistics has these numbers, which suggests it's a big problem. So yeah, if we just did XYZ, then introduce technology to the medical practice, all the problem would go away. And here's a really enormous TAM number associated with it. So I guess there's a cautionary tale there for other young founders. Although I don't regret it at all, I think.

9:54Op-Kit was my opportunity to get started on my founder journey. YC helped me, you know, helped kick me out of the nest, so to speak. I really feel like a founder. I think that this is the right career for me. And I was looking for an invitation to get started. And I think Op-Kit and YC gave me that. And then now years later, you know, after a mediocre outcome with Op-Kit, I feel like...

10:21I've learned so many lessons that I get to now apply to my current company. And the only reason my current company is as well positioned as it is today is because of the four years we spent eating crap for my previous company. And so these learnings compound, the relationships compound. It's not a finite game. I don't know if I had a 40-page memo associated with op-get. Maybe we would have benefited.

10:51from having to doing a 40 page memo. Interesting counterfactual though, right? It actually would have been easier for me to write. It's easier for me to write it for Sizzabi when I actually know all the answers already than it would have been for this other idea. Maybe the process of trying to write that memo would have been what led us not to do it. Exactly. Yeah, that's what I was thinking. So going back on this point of you already know what you want to write, talk a little bit about how you got there, right? So, you know, you're at Brex, walk us through, you know, I love these stories of like, hey, I almost rode Rage Quid on this day, right? Because this thing happened, if you have any of those stories, right? I'm sure you just had these moments when you were like, wait a minute, this has to end. Or maybe not, maybe it was an accumulation of things, but walk us through that.

11:43I guess it would be useful to explain how I got into infrastructure to begin with, because I have an untraditional engineering background. I studied history at Davidson College in North Carolina, which is famous only because that's where Steph Curry went to school. I'm certainly not famous for programmers. You could change that. I don't even have a CS undergrad from Davidson. They didn't have a CS program while I was there. They didn't have one. They had probably two or three courses. But nothing proper. Nothing proper.

12:12But I took those first two classes, and I was like, holy crap, this thing, this is so fun. And I already love computers already, and I'm getting to make them do interesting things. There's more than I could do than just download music illegally or mine Bitcoin or my remote desktop.

12:34It was a revelation for me. And then what really led to a takeoff was discovering Web Development Bootcamp in San Francisco called Dev Bootcamp, which I think was the first in-person Web Development Bootcamp. I'm sure that there are many programs that claim that honor. But I think it was Dev Bootcamp. And what year was this? I must have read, so I discovered it through a Forbes article.

13:01I must have discovered that article maybe in early 2014 because I went to the program in the summer of 2014. we were the 2014 chorus frogs and all of my peers are like in their late 20s early 30s like teachers or dentists and they wanted to become software engineers. That was like the only undergrad. I'm just here for the summer. And is that how you met Rex who Rex Salisbury who also went to Davidson? Not originally how I met him. So Rex was class of 2010 at Davidson that was class of 2016 so we did not overlap. Rex is now like one of the only and most successful David's alumni in tech in San Francisco. I stayed on his sofa for like a month when I first moved out here. He did Dev Bootcamp too. I think he did it a couple of months after I did, but because he was, I think at that point, he was a couple of years into his career. He was transitioning from an actual banking role to an engineering role at a fintech company.

14:01It's like what he's known for today. Yeah, I mean, he's a great investor. We've done a number of deals with him. Yeah, he rocks. He said he's obviously invested in Cezabi. And he's said repeatedly that DevTools are not his thing. But if it's working, he doesn't mind. I mean, it's like most of tech, right? Or most of VC right there. Yeah, yeah. It's like you'll take a W where you can get one. That's right. That's right. OK, so Dev Bootcamp 2014, you're surrounded by folks who are different.

14:29than you. What was that part like? Did you have to do group projects? Absolutely, yeah. I think Dev Bootcamp was the first time I'd been to California, first time I'd been to San Francisco, first time I'd probably done yoga because they would do yoga at lunch breaks. It was a really hippie program. And you can imagine, I'm a Southern white frat boy. What is this?

14:57But I also never felt like I really belonged in that culture and I loved it. You're open to it. It sounds like yeah I was really excited about it. I think I've been looking for something different in California and tech was something different Yeah, and I felt like I was really good at it. It was very creative You know, I'd spent a lot of my adolescence like making music and some of that music was digital like digital audio production so suddenly it's like completely expressive. I can use computers to make all of these apps and it's not just like little snake games and a Python interpreter. It's like a full app and it's got a YouTube integration and I remember that summer Node came out. I think it was, I think Node came out. It might have, maybe it was like version 1.0 or like an early LTS but like server-side JavaScript was new. We were learning Ruby on Rails at the time. So it gives you a sense of...

15:50It's what Airbnb was built on. Yeah. Obviously, some people still love Brails. Shout out to DHH. That's right. But it was a revelation for me. And from that moment onwards, I was probably 18. I realized I want to come work in engineering role at tech startups in California. And I want to do my own YC Ventureback tech startup. And then I continued to chase that through the rest of my undergrad. It led me to crunch base after graduation.

16:19And as the junior-most engineer at CrunchBase, you don't really get to work on any interesting features. So the one thing that was kind of left over for me to work on was our CI-CD pipeline. And then, as you can imagine, here I am, infrastructure engineer, however many years later, well, 10 years later. That's how it happened. Wow. There are so many things to dig into there. One that I want to circle back on is a comment you made about programming being creative, right? VC is intact. It's so full of hot takes and binaries, right? And one of those binaries, I think for many years was, you know, we saw this ascendance of stem degrees. And it was basically like, if you're not doing a stem degree, then you're probably, you know, you're making a terrible choice. And what's your future gonna be gonna look like? You studied history, as you mentioned. I feel like we've kind of swung back now to thinking more about history, thinking more about the role of creativity in what we build. There's a lot on that bone. Yeah. Well, whatever you want to chew off. Yeah. Where should I start? I mean...

17:32I won't claim that a history degree is superior for engineers or for startups. I don't know if that's true. Certainly not for engineers. You could say that I spent my entire career peeling back.

17:47Layers like I think started at web development wanting to go further closer to the metal and I really kind of stopped at like Linux and and Infrastructure because that's where the market demands it and became a founder around that time, so I'm not racking servers I'll be up forthcoming about that, but I've spent you know last 10 years trying to learn the next layer beneath me and I think, you know, that's not how everyone learns engineering, but for me it really satisfied my curiosity because I'm kind of start with the layer of abstraction that allows me to do something cool and then I want to do something a little bit cooler. And I'm like, well, the only way I can do that is apparently is I have to like learn this next layer. And then you start operating at both of those layers and you now have more.

18:31creative flexibility or leverage and you can build something more impressive and then you run into a new barrier and you have to peel back the next layer. I think that programming is really creative and expressive. I'm a very visual person and I think in terms of systems and abstractions and shapes and object hierarchy, abstract and instantiated classes, systems of rules.

19:02I think it's a very different way probably than thinking about algorithms and data structures. In fact, sometimes algorithms and data structures can be hard for me to grasp. But when we talk about system design, or the design of a large code base or what's axiomatic about a code basic? What makes this, why does this file live in this directory as opposed to this directory? What are the truth things that must be true about this file in order for it to live here instead of somewhere else? Those things come really intuitively and it feels like kind of like molding with clay.

19:36Yeah, it makes sense that you describe it as creative. Then how does that show up in how you build your team? Who you hire? I've met Justin a couple times. He's fantastic. Yeah, what's that relationship like for you all? Is he more capable? Isn't the right word? But does he thrive more and more?

19:59Algorithmic tasks are interesting. I don't know if I would say that that like we that's no those aren't the dimensions that we are complimentary on And Justin is my you know former co-founder from first company now founding engineer at Cezabi Best friend roommate collaborator at four different companies. I think well We go way back And he's an amazing compliment to me. I really could not be building Cezabi without him one of the ways that we're very Complementary is that I tend to be extroverted and I tend to verbalize as I think Justin is more introverted and more thoughtful and reserved feels like it's doing a disservice He's not a shy guy, but he will you know, he'll think about something before he shares what he will come to conclusion before he shares that I like to think throughout ideas by verbalizing them so you often hear me like try on an idea by speaking and and then

20:57I'll be like, I'll walk it back and say, actually, that's not quite what I meant. Which maybe is not well suited for podcasts, I'm not sure. But in that way, I think he and I collaborate super well. And we work well on problems. It's a nice counterbalance. Does it manifest differently in, this is kind of silly, but even in like PRs, do you kind of fire off? I mean, I know agents have changed all of this, right?

21:26You know, some people have like really stringent standards around like commit messages and, you know, they want to give all the context. And I mean, you're working at a small organization.

21:36Organization now compared to Brex, but yeah, do you see it manifesting like am I firing off PRs and Justin's like, you know, mine are like poorly tested I don't think there's a correlation there But the way that we solve problems together as a team is that he will let me kind of generate ideas and Basically spitball and then he'll help reel me in or I'll help like prod or or steal man some of the ideas that I come out with. And it usually helps me come to a conclusion, and Justin will be sitting there like he knew the conclusion the whole time. Perfect partnership. It's great. It's a great team. We need people who I think will let us explore. There is, I think, often this inner drive to be heard or to go through that journey. So it's great that you have a partner who can help there. Yeah, I mean, without the right thought partner, I think that my advantages aren't...

22:31Very advantageous like it's not not very useful to be just generative without the feedback And so like you need both sides. Yeah. Yeah Okay, let's jump to actually talking about sissabi a little bit. It's a really exciting time for you You just started YC for the second time Which I'm I'm thrilled about in that you know, I can't wait to see I mean, I'm already seeing the momentum, right? Just seeing the product continue to grow, obviously, customers and revenue coming for that. But let's take a step back and talk about the promises of Sizzabi. We could talk a little bit about the manifesto in terms of, like, there are a couple of core claims that drive it.

23:21And that are a little controversial. So let's spend some time there. I mean, just the first logs are all you need for folks listening who don't know about the space, frame that in terms of the space, and also talk about why it rubs some people the wrong way. Yeah. Well, for now, I think a lot of the core ideas behind Zazabi become less and less controversial the more that AI plays out. Because they seem more self-evident. I mean, they seemed self-evident to me a year ago when I was writing about them. And now they're seeming increasingly like obvious, which is great. I mean, it's very validating. I'm glad we got started when we did. I mean, for context, Adobe is the AI native observability platform for fast-moving engineering teams. It's effectively like Datadog, but rebuilt in 2026 instead of 2010. Yeah, and walk us through what that

24:13I have an innate sense. It's almost repulsion. Strong word, right? When you say that, I physically know what that means. There's a physical thing. I actually much credit to Datadog. I think for a long time, Datadog set the standard for product quality in the space. They kind of defined observability, or the observability era, in the 2010s. Certainly defined the transition to cloud-based monitoring.

24:41helped us through this microservices period, which I think we're mostly past now. I saw you tweeted about that today. Twitter I just have a lot of fun with. But I think Datadog is still a loved product. But we also really liked VS Code before Cursor came out. I remember thinking VS Code had everything I needed. So there's a better way.

25:08And there's been a technology paradigm shift. And I think there's an ongoing paradigm shift in the way that we build software. And I think that the old way of finding and fixing issues in your production systems isn't just not going to continue to be the norm. And walk people through that, right? I'm thinking metrics, dashboards, describe some of...

25:35I guess some of the pain that you're responding to here. The three big ideas behind Cezabi are less is more, logs are all you need, and traditional monitoring is dead. And they was kind of point to, I think, the challenges that are associated with traditional monitoring, which I think all existed prior to AI and prior to Cezabi, but were exacerbated by AI. Less is more basically means that there should be less UI.

26:03If you open up Datalog today, you're looking at tens, maybe 100 modules. Obviously, all the items in the sidebar, all of the sub-items in each of those menus. And then within each one of those screens, hundreds of configuration knobs and buttons. So it's a complete overwhelm in terms of functionality and user interface. But it also means less.

26:30like less complexity, this flows into our second hot take, which is logs are all you need, but less instrumentation. This idea of logs are all you need basically refers to the fact that in contrast with the three pillars of observability idea, which has been dominant for a long time, we believe you only need to provide logs, not metrics or traces.

26:57It's not to say that instrumenting metrics and traces on your services can't be valuable, but just that it's unnecessary and superfluous. This is a very controversial idea because for a long time in observability, the three pillars, if the first idea of observability is that, observability is about asking new questions about your system, things that you didn't anticipate, then the second big idea is the three pillars. Everyone just cargo-acculted this idea.

27:24Observability vendors like Datadog have been, I mean, not to say Datadog or Observability is a Psyop, but they've certainly benefited from the idea that you should be instrumenting more and sending more data and that you should have three modules or three sets of modules instead of just one. So you better believe that they lashed on to this idea and then it let it define their product roadmap and then sold it back to you.

27:53I think that logs satisfy all of the use cases that metrics and traces do. The original insight came from this, it was from my personal experience. We had, at many companies now, I've bought and configured Datadog and configured all of the modules, so all of it was available. But when something goes wrong, the first thing I always reach for is the log screen, and then I start doing log queries.

28:22Just at some point occurred to me like I'm not using these other pages and There's a huge amount of overhead associated with setting them up and maintaining them And then that happened simultaneous with this realization that While logs used to be something that were you know disadvantaged because they were unstructured You couldn't have a machine read them at scale and then there was a constant push to make your log structured like it's It was like at every organization I've ever worked at it's like we've got an ongoing migration from unstructured logs to structured logs and it's Happening in perpetuity and who knows like when this migration will be complete AI is now able to read that data at scale in a way like that wasn't previously possible And those logs are really rich. I mean the information that's contained there like the decision to

29:18Like the way that you write the error message like whether it's like this is a really bad error Or like this is like this seems suspicious like that has semantic meaning and value on it and AI can actually use that And then the last I mean I'm getting on a soapbox here with regards to logs or all you need the last point I'll make here is that Logs are essentially the same substrate as metrics and traces a log is an event And a metric is just an aggregated event. And a trace is a collection of start and end events. So if you have the logs, you can reconstruct the metrics and the traces. And now you've only asked your customer, the user, to instrument standard out and standard error, which is way easier than stats D or dog stats D or Prometheus or Jaeger or trace propagation.

30:11It's way less failure-prone. It's like the Occam's razor of observability. So it's just a better product experience and user experience. So, yeah, there's so much to dig into here. I'm reminded of, you know, is it trained as a data, well, trained? Oh, yeah, this must fly in the face of many of the principles that you know. I mean, I spend so much time with logs, right, at Airbnb. And, you know, there are so many challenges with, again, pre-AI.

30:40So much of what we're doing with logs was across device logs and inconsistent practices across product teams. I mean, I'm curious to get your thoughts there. I actually haven't dug too deep into are there any tools that are making the collection of logs at scale easier? That's a great question. I mean, there are things like vector. And honestly, I don't think that there's a ton of innovation in this area. Most people are still using open telemetry to log in libraries or log forwarders or log shippers. There's probably some interesting work to be done on the client side with regards to proactively filtering or aggregating logs that might reduce the payload size or reduce the size of the amount of data that's stored.

31:36ultimately in the end or Or just reduce the amount of network like network traffic or bandwidth that's consumed I don't think that that's where the value is. Yeah, I think AI is like allows you to do fuzzy everything and What's great about that is that like it's okay if you're you know, you have one service that emits the service name with a, you know, in-sentence case. And another service that does it in snake case. Like AI can resolve those two things for you. You know, in the past, I think that we would, we'd have solved these issues by enforcing standards on teams. There would be like an observability team. And I know that because I've been in this team, where you go out to all the other teams and you kind of slap them on the rest. And you're like, Hey man, didn't you read the guidelines for how to instrument your services? Like, or.

32:24Are you instrumenting using your own framework instead of the framework that we built for you? And it's messing up all of our telemetry and like now you can't correlate properly. We don't need to live in that world anymore. We can have AI paper over these minor differences.

32:44take on so much burden as product engineering teams, which is, it really is, there's just too much burden into the expectations for product engineering teams who have completely different incentives, completely different priorities are too stringent. Yeah. Well, and it's fascinating too, to, you know, as I'm reflecting on what you're saying, you know, in many ways it's like, in situations where we don't have control, just like analyzing a little bit, right? A situation where we don't have to control. We often try to take more control, right? And I think metrics, traces, all of this is a really fascinating manifestation of that, right? Because it's like, oh my gosh, so much is happening between my services. No one ever really has a through line, right? And as much as we tried to create that, right, we tried to create it by just instrumenting more.

33:37You better believe the observability vendors love that, too. They're like, of course, like if you know the solution to your problems. Just one more. One more logging, one more log. And actually that's a question, like sometimes people or skeptics will ask, you know, what if I don't have all of the logs that I need. And I think we've been really surprised when you use Sizzabi, you connect your code base as well. So Sizzabi has access to both your log data and to the source code.

34:06We've been very surprised and impressed with this ability to basically read between the lines. So if it has a log stream that goes through a code path that isn't fully instrumented, like you're missing log lines in certain code paths, it can assume or infer what has happened based on the source code. And it's usually right.

34:29We have other mechanisms where we try to encourage people to instrument more logs. We're working on a GitHub action that will add logs automatically to your pull requests, or using the CLI to basically scan your code base and give you some score about which parts of the code base are code coverage, which parts of the code base are well-instrumented versus not. But in general, I've been impressed.

34:55the agent's ability to infer what's going on, even in the absence of telemetry. That alone is ... I don't think we can fully wrap our minds around that. I follow some of the interpretability research that Anthropic, as an example, has put out. Hearing you say, you're obviously not the only one, who's like, oh, we noticed this thing, it can connect the dots. It can connect the dots in a way that ...

35:22actually is correct. Like a great staff engineer would do the same, right? They're like, I see that the log stream kind of stops here. But I know that what change recently went out, I can connect the dots there and draw a conclusion about what probably happened. And in many ways, that is the service that you're providing customers, right? I mean, I know you don't use this marketing language, but I think so many companies what they're selling is it's like just imagine right you have this it's almost like the rubber ducky problem in a way like I when I think about my time at Airbnb there was one dad engineer who could answer all of my questions all the time and I valued my time with him so much that I'd like save up my questions you know so we could sit down and I could like go through each of them and now we get that on demand which is pretty at scale right for every member of your team for every member of your team

36:16Yep, your executives can talk to that person as much as they want to. Are you seeing that yet with Sizzabi? I mean, I'm an executive and I use Sizzabi. That counts dog foodings, you know? Incredible. I think we sell into CTOs and engineering teams. We have some sort of clever tricks to promote, like adoption from other parts of the organization.

36:38if a Slack user who isn't already signed up to your organization, but is a part of your workspace, mentions Sizzabi and asks Sizzabi a question, they automatically get provisioned an account. So it's designed to spread within an org that way, and certainly to be used by operation and support product. Even for executive use case, I can just be like, what new features do we need? Or how is this high value account doing?

37:08You can use it for product analytics, basically. I find that we use it for all sorts of use cases internally. Although I go to market and position it as a relative to Datadog, because that's our main focus. The buyers are the same. Exactly. We're finding that it's serving a lot of use cases beyond that. That is the beauty of the chat experience that we've all...

37:35Reverted to but there's something comforting in it right there is something kind of natural to chat you know back to ICQ or even you know Google chat back back in the day. There's some other correlaries to this like for example One point of controversy we often run into is dashboards because people want them dashboards Yeah Personally I think that dashboards are dumb Tell us how you really feel. Yeah, so we're just airing it all out. I think dashboards are mostly about vanity or they're about psychological safety So you either want that dashboard because you like it makes you feel good when you look at it Or I guess I mean this is also the sort of the same or you are nervous about the stability or performance of your application And the thing that alleviates that nervousness is like putting it on a screen so you can look at it, you know at all times I don't think that people find

38:34issues, like new problems from dashboards generally. And I think the, an ideal in-state is like you assume the product is working and you, like your default assumption is operational and you only, like it's only add down when you get alerted when there's something broken. So in that.

38:59in-state, which I think is the logical conclusion with all of this. Like, you don't have a dashboard, you just have a staff engineer, like Sizzabi, reach out to you and let you know, like, hey, there's a problem, this customer's affected, and this is what it is, and this is how you resolve it. But it's a tension that we're running into right now with the product where we're asking people to give up something that they've...

39:18Yeah, these things that they've curated and managed. Yeah. And it's given them comfort. Like you said, like, oh, I'm looking at the chart, right? Yeah. Yeah. There's a tension between the world we think that people should live in and the world that they live in today and the Overton window and how we move from point A to point B. Yeah. Does it show up in our product roadmap? Yeah. Yeah, it's interesting. I mean, when we first started chatting a couple of years ago at this rate, I probably talked to you about my thesis, for lack of a better term, on headless BI. So again, I was trained as a data scientist. And exactly as you noted, in many ways, my job was to make dashboards. And I was like, as soon as you make a dashboard, it's over, right? And so much of it was just doing that over and over and over again. And then you make it, and someone looks at it and never comes up again. Yeah. I mean, you're wearing the hex.

40:12sweatshirt right now. I am. I think HEX is kind of solving this for data analytics. Yeah. And they're, in many cases, you know, like Sazabi, they've always been ahead of the curve, right, in terms of, I think, taking product risks where other people wouldn't, right? Which is, you know, it's always great to have those examples. But you know, I had this thesis, which is panning out, I guess, in various ways around, like, don't make me do the work, right? Like, don't create a product where it's my job to, you know, go to a UI. It's my job to do the thinking in some ways, right? When all I'm doing in many ways, so much of dashboarding work is about anomalies, right? It's anomaly detection. You know, maybe you set some parameters or something, but it's like...

41:03That's how we did it 10 years ago. We're so far beyond this now. And just this idea that we should be meeting users where they are, which is often in Slack, right? It can be on WhatsApp, iMessage, whatever. For me, it feels very natural.

41:22Again, don't make me do the work. You have all the data, right? You have access to my code base. I've set thresholds come to me. And I'm really intrigued by this comment you made that the market might not be fully there yet.

41:38I'm curious, where are you seeing folks get really excited about it? Are you noticing, are there other trends that you see with these customers who just lack of a better term? Are they AI native, where it's like they're used to this and other facets of their work? Yeah, do you see any trends so far?

41:56That's a great question. We focus on selling to Series A and Series B companies. We find that earlier-stage companies tend to be more AI-native and tend to have more appetite for AI tools and for novel approaches. I think these companies have fewer established tools and processes. I think they're also just looking for alpha. They want to find a way to accelerate. One of the things that feel slow right now is maintaining production and fixing bugs.

42:26So, talk about the self-healing aspect of this, right? Because that's on your roadmap as well. That would be a huge unlock. Just through what the vision is there and what it looks like in practice. Yeah, it will be a very strange world, I think, when it becomes widespread. But Sazabi's mission...

42:47is to be the default observability solution for companies of all sizes during the AI era, which we define as 2025 and beyond, so the year of agents and beyond. And our vision is a world where software heals and improves itself without human intervention or direction, which basically is self-healing software. You can tell I've seen it. Not ambitious at all. Yeah. I think in that future, software looks kind of organic.

43:15Self-feeling can probably be defined as a software that changes itself in response to some stimulus in production, probably like an outage or a bug or an error. You could argue that we've already had self-feeling software in the sense that we have automatic rollbacks. Some teams have automatic rollbacks.

43:37Still mostly like a holy grail. Kind of like preview environments. You should be so lucky that you work in a company that has one of those. That is a pretty weak and uncompelling version of self-feeling software and it doesn't do justice to what we want to build. If you think about like nature...

43:56When things heal it doesn't always mean we're returning to a previous state like a bone breaks and it heals stronger Than it you know, then it was previous or you know, you repeatedly strain a muscle over your lifetime It becomes developed or over generations becomes even more developed I think that the world of self-feeling software will look like that where software changes and evolves and becomes more robust and maybe it even develops new characteristics or features, mutations, if you will, in response to production usage.

44:30That's the world that we want to build with Sizzabi. I think requires access to production data and the ability to interpret and analyze that production data and feed it back into the rest of your software development lifecycle. So I think Sizzabi kind of represents the last link that allows us to create that. How much are you thinking about the next model releases, right, in terms of your product roadmap, in terms of that?

44:57Vision, right? I mean, I don't know if you're thinking in terms of you know if you have a time frame and and and it's already taking this into account I mean none of us obviously know actually what's gonna happen, right? But it is attention right like you have to build for where we are today even if in six months the world in terms of What Cezabi and all of us as builders are capable of doing is not pretty cool. Yeah, I mean it's an it's an anxiety inducing. Prospect, I think I was thoroughly AGI-pilled, whatever that means, when I initially came up with a premise versus Abbey, and was totally convinced that all of these little bells and whistles, like dashboards, flame graphs, service maps, even metrics and traces, were superfluous, and that in a world where we have cheap, abundant, and performant,

45:54AI models, we won't need them. And so that's what led to our key features, which are like the idea that you only debug things. You don't look at your telemetry directly. You only debug things through conversation. You just ask an agent what's wrong. And the idea that you can't create monitors, instead you just receive like agentic alerts or autonomous alerts.

46:21which are entirely agentic in the sense that the agent is looking for anomalies, investigating them, deciding whether to escalate them to you. There's no way for you to build in a particular learning condition. Now, there's an asterisk there, which is that you can tell the agent to look for issues in particular parts of your system or to have a higher sensitivity. So maybe you have a payment service, I can say.

46:48Anytime anything weird happens in payment service, make sure to let me know. And that will change some of the learning behavior. So I want to get to brand, because I think that what you all are doing is pretty special and fun. But before that, going back to what you were just saying, are there code bases, challenges, I guess you name it, that Cezabi is particularly well positioned to you know fix, solve, self-heal. Do you have a point of view on that? Yeah, that's a great question. I would say we're kind of discovering that as we go.

47:32I wish I had a more satisfying answer than this. It's a little bit vibes-based. In the same way that I think people are discovering what types of code bases and technologies and genetic coding works well for, we're also discovering what types of code bases and technologies and tech stacks and service architectures are easily for us to investigate and monitor root cons. We intentionally seek out customers that have Weird stacks or that build like maybe this customer is building a piece of infrastructure Like hosted Postgres or something like that that we haven't really like served at large scale before and we want to see how Suzabi performs on that use case so We don't have I wouldn't say that there are specific stacks that we think the Suzabi works well for right now This also feels like a little bit like it's it's not immaterial because again in this in the

48:26If you can assume a certain rate of model improvement, then you can assume that for any code base, any tech stack, any service architecture, this is obviously going to be effective at root cause analysis and in detection. And we do assume the models are going to continue getting better. I think it's a basic premise, probably, of what we're trying to build. We're trying to build observability for late stage AI. Yeah, super powerful. OK, brand.

48:53You all launched your website, what, a week or two ago? And it's honestly unlike any other that I've seen. For better or worse. Yeah, I mean, it's bold. Yeah, talk about your website. Yeah. And maybe we can get to the name a little bit as well, right? I'm always learning new things from you. I obviously had to look up.

49:20What's the Zobby meant before I invested? Really critical diligence point right there. We're going to out me as a massive nerd. Exactly. We started there, so it's fine.

49:33I guess the website is a sort of an immersive experience where you are in a kind of like a mech suit or or like a pilot seat and you're looking out over over like a city during nighttime and on different buildings of in the city there are indicators that indicate issues with the city. So there will be a red light that says there's some 500 errors here. There's a green light that says this component is operational. And there's a yellow light saying there's a latency trend that looks concerning. And you can interact with them. And you'll see the dialogue between yourself, the operator, and Sazabi, root causing them. And you get to click on little buttons to help proceed to the next phase.

50:18And then below that, we have a more conventional website, but I will say that I think the brand really shines through in the rest of the website. The textures that we use, the bezels, the illustrations, the graphics, the typography. We have scan lines across most of the site, which gives it this kind of like...

50:40Sort of like a blade runner yeah vibe and and like a sense of texture that other websites don't have on some of our header typography We have like a subtle glow or gradient inside of the the typography typography itself and did you drive most of that you work with a like where did that come from? I've worked with two studios on this on this project. I worked with one main studio and I think When I brought the initial concept of the brand to them, they loved it. These guys are just equally nerdy, I guess. And they really ran with it. And I think, you know, I have to give them a ton of credit for bringing it to life. Yeah, I've created mood boards. I created early versions of the brand on my own that I brought to them. And I think that there were specific.

51:26visual elements that I loved and wanted to incorporate. A lot of them come from video games. The glow around the typography is something that I saw in Call of Duty Black Ops, which I thought looked really cool. Black Ops also uses layered textures and illustrations, which are really cool in some of the backgrounds. It's very maximless in a way that I think is different counter to a lot of design today.

51:55it's it's a it's uh It does not it's not consistent with sort of the Shazian slot world that we will live in today or the post linear design I don't know. I'm not sure what to call it like it's like authoritarian design worlds You know where everything has to be clean and minimal say more about why that's important to you not that aspect in particular But the whole project right yeah, I mean it's Zobby's a personal passion project for me like I had one investor describe it as my tombstone company and I thought that was an amazing way of describing it, because this is the company I want to be known for. I don't plan to start another one. I think I called it Sahzabi because when I...

52:36Set out to start another company one of the first things I wanted to make sure was true was that it was fun Mm-hmm. I grew up a massive mobile suit Gundam fan, and I continue to be and in mobile suit Gundam, which is this 1980s sci-fi anime franchise that there is a mecca mobile suit that the antagonist pilots called the sozombie and it's super really cool red Futuristic suit. It's very powerful and fast And I thought, that's just cool. That thing makes me go, hell yeah. And I want like, there's no observability tool that makes me go, hell yeah. I want to build a developer tool that makes people feel that way. And I wanted it to be fun. And so it was almost a symbolic act was to choose to call it or name it after something that I love. Yeah. I mean, it's bold in the best way. And I think looking, thinking back to something you said a bit ago around

53:31You know tech engineering is creative, right? I don't think it is for all people it clearly is for you right in terms of what you bring to it Why you bring what you bring to it? What have been other developers reactions? It's divisive. Yeah, you know I mean I think starting with a name some people love it some people hate it it sounds like wasabi I've gotten into the habit of spelling Suzabi salsa, you know, I work at Suzabi, S-A-Z-A-B-I, which is totally fine. We have our dot-com demand, so I'm happy. We have all our OG handles on different platforms. I'm not too concerned about that, you know, Shakespeare said what's in the name, you know, like, they're great, you know, bring it back to music, like, so many great bands have ridiculous names, Coldplay, Radiohead, like, they're great because of the, you know, of demands. You can remember them, right? You know, and I'm certain that like, Suzabi will be the same.

54:25The brand is also divisive, I think. It's very red. And red makes people feel a certain way. I'd much rather break through and to have an opinion than to not break through at all. And I think we're much more likely to attract a following or for people to be excited about using Sazabi or working with us or about being associated with our team or our brand if we are.

54:54if we have an opinion and a voice. And so I'm not concerned about that. So did it feel, was it an active choice on your part, meaning? Obviously it was an active choice, but do you feel pressure to shift anything? You know buy investors by No, I mean I guess we do focus on serving like startups like early-stage and mid-market like high-growth startups The main reason for that is because you know, I've worked at high-growth startups my entire career I think the largest company I've ever worked at was probably 500 employees, which is Brex And I just wanted to build I wanted to build a tool for people like me or for engineers who worked at those companies. Those are the types of companies I could confidently sell into. I understand their priorities. It makes sense for me as a target or initial target market. I also think that a lot of new AI observably focused companies are focused on enterprise. Datadog is very definitively focused on enterprise. So there's this nice opportunity to serve this segment. Startups, I think, are...

55:56If I were selling it to Enterprise, we might need to have a little bit of a different brand. Fortunately, we're not. We're selling to startups. And I think Hex has an amazing brand. Another shout out for their team. But it's also different. It's one of the luxuries, I guess, of serving this particular market. Yeah. And as an investor, you know this about me. But in general, I'm so...

56:23obsessed with slash one of the most important things for me is just finding people the point of view right a point of view these days is harder than than anything else right like as you noted the the slop you know the information environment is just full of you know, regurgitated thought, sloppy writing. And so, you know, I do feel like a big part of my job is cutting through and finding folks who have a strong, you know, opinion about a product that needs to be made, that needs to exist in the world, you know, a problem that keeps them up at night. And it's because I primarily invest in data dev tools, you know, like you said, it's even rarer to find that.

57:04a POV shows up there. So it's a welcome change. Going a little deeper into the developers that you sell to, you have had a really interesting angel investor strategy. I think it's one that many people could not pull off, but I think they strive to.

57:24So, just talk a little bit about what the angel strategy is, how you executed on it, and what brought it to mind is actually the part of your website where you actually, the previous iteration where you had folks in Tony Stark suits. Yeah, we wanted to do that again for this version. We had to cut some scope. I guess, to share context, we have over a hundred.

57:51angel investors on our cap table. They're all on roll-up vehicles, so future investors can feel comfortable knowing that our cap table looks quite clean. But these people are typically founders, CTO, CEOs, technical leaders, at basically the best AI and developer tools startups in the world. Companies like Browserbase, Replet, Versailles, Daytona, Graphite.

58:20Many others the strategy here is that you know, I I'm a big believer in like creating me We're here. We are a village global sort of creating a village. I want as many allies as possible I want people to be bought into my success and to the company's success and to take them along for the journey some of the investors in my and so Zabi are investors that I worked with on my previous company I think like the angel investors are a way for me to like consummate That relationship and to say like hey, you know, you're a part of the team You know like we're gonna this is obviously gonna succeed You know with with you and because of you and you'll be a part of that story It's not as transactional as me thinking like Because you work at this company and we will get a pilot at that company although in some cases it does play out that way Sure, you know people move companies. Yeah, so that we don't always get to we executed on it in that fashion, but certainly that happens I think

59:17You know, early on, it was just me and a long memo and a vision and some strong opinions. And I wanted to basically sell this idea and this vision in the absence of a team or a product. And one of the ways I could sell to the companies that are my dream accounts before I even have anything in hand is to sell them on the vision and on me, the founder, and to bring them in as angel investors. And I think it's created opportunities for us.

59:46Who are the best angel investors? You can give their names, or how do they show up? So we get lots of founders who listen to this content that we produce, and they're always looking for advice, right? Well, one piece of advice I'll share is I generally only take money from people I've met in person. I think there's a lot of money in Silicon Valley. And if you're a sought after company, you're going to have lots of people who want a piece of the cap table.

1:00:15I think you should like your investors and you should also be responsible about diligence in them because you want to make sure that you're in bed with the right people, people who share your values. So that's definitely a recommendation I would make to everyone. In terms of best angel investors, I love having someone I can text about anything. I've got 100 people on my cap table.

1:00:43Someone who's an operations leader or someone who's a founder of a post exit dev tools company or someone who's you know series a company So they're just two steps ahead of where I am or someone who is in more an infra and someone who is Like a buyer at a another tech company that isn't an infra But when you have this many angels you have someone you can ask for a favor or for advice that you know in any position Yeah, that's really nice to have that and you know ideally like These relationships are close enough where you have each other's phone numbers. You're going to happy hours together. And these people become friends and general supporters. It's obviously great when you've got angel investors who are actively trying to bring you into accounts or trying to promote you. That's not my expectation.

1:01:35Maybe one of the big things I learned from being a second-time founder is that you should like your investors and you shouldn't hesitate about asking for favors. I think a number of times I've asked you for things. Yeah, no, you've graded it. Consistently come through. But as a first-time founder who's a little bit younger, less experienced, you may be a little bit more shy about reaching out to your investor. You should make them work for you. Definitely. Yeah, why aren't they there? Otherwise. Make them record podcasts with you. Yeah, there you go. What is the most annoying thing about selling to developers?

1:02:04The most annoying thing about selling to developers, I mean, developers are such skeptics. For the people who occupy the very beginning of the adoption curve, they're still so skeptical. They also think that they're immune to branding and marketing. They're totally not. Say more. Well, I mean, maybe we'll find it. Maybe Suzabi's brand will just totally flop it. We'll discover it. They're right, but I don't think they are.

1:02:33I'm I certainly am not you know I want to use developer tools that make me that make me feel cool. I Think developers hate to pay for things until they've seen value Which makes like anything that sales land really hard just kind of have to do product led Which has a whole bunch of strings attached and considerations for how you build your product how you build your business how you know market how you deal with competition. Any lessons you can share there that you've learned so far in terms of time to value? Are there any, like, aha moments that you've had processes that you've changed? Naturally, we could get into onboarding. But yeah, curious to hear your thoughts here. Yeah. I mean, Cizabi requires you to connect a few different things before we could provide the most of our value.

1:03:23For example, we ask you to connect your code base, like a GitHub connection. We ask you to install the Slack app, so you can communicate with Sazabi over Slack. That's required. And receive apps. It's not required, but you're going to get a lot more value if you take those two steps. And then last but not least, we ask you to send us your production logs. So we're asking a lot before you. Logs are not all you need. I mean, you've got to have some integrations, too. Sure.

1:03:49I think one of the things we've discovered over developing the product and rolling it out to some early customers is there needs to be some trust and value delivered early. One of the ways that we've solved for this is...

1:04:02you can use Cesabi with what we call an external backends, meaning you can have Cesabi's agent and alerting system operate on top of a third-party logging storage system, like Datadog, or Axiom, or Clickhouse, or Sentry, or Postog, or any one of many other logging backends. And that...

1:04:22What that does is just lower the barrier for adoption. You don't need to actually send us any data directly. You can just plug in an API key and start seeing what it's like to use Cesabi and getting some of the value. Of course, Cesabi is way more powerful when you use our logging storage. It's also cheaper because you wouldn't be paying for data log log storage while also paying for Cesabi inference. So I would say that we've had to discover an implement.

1:04:50little features like that in order to unlock value at earlier points in the customer journey for developers. And then meanwhile, we try to support both product-led and sales-led motions. When we approach later-stage customers and it's more of a sales-led motion, the bar for adopting a new observability solution is quite high because this is something that needs to be...

1:05:16We're very reliable, needs to be available and useful in the event of a production outage, which is like the most, like this is like your, if you're bleeding out, it's your support tool. And so that I think the requirements for adoption there for later stage customers are pretty high. So one thing that I think, It gives me peace of mind is that Suzabi's had to invest a lot in infrastructure security and compliance. And that's not something you can vibe code. So for now. Okay, so Facebook had their seven friends. It's basically Canon in tech. The idea was once you get to seven friends, you're basically hooked. You're not going to turn from the platform. How do you think about that at Suzabi?

1:06:08Is there some sort of network effect or viral? Well, more like, is there something that Cezabi can tell? Yeah. That is a great question. I think once you start receiving autonomous alerts.

1:06:24it becomes clear how powerful this is obvious and how useful it is. Because just by connecting these things, not by doing any configuration or setup in the platform, like you said, we're bringing the value to you instead of expecting you to buy the platform and then do hours worth of investment. Once you start getting those alerts, I think that's usually when people have an aha moment. Oh my gosh. I've put in close to zero effort.

1:06:50Higher signal more actionable richer alerts than I've ever had at any previous software developer role in the past I think generally like we focus on the connection of Slack GitHub and Having production logs flowing because those are the three things that like once they're in place like you're really getting maximum value Yeah, I mean it's it's been on Slack in particular. It's been wild. And I guess it's not at all surprising, but at Airbnb we were very early users of Slack. And I was new to tech. Airbnb was my first tech job. I had come from finishing a PhD. It's kind of nothing further from tech. And I just remember Slack just being like, what? I'm supposed to show up here and just have conversations. It just was...

1:07:40I couldn't wrap my mind around it, but we were such avid users that obviously I got on board. But, you know, fast forward then to today where, you know, it's literally where work is getting done. And even when I go on vacation, I can't, like, take slack off my phone, right? It is the lifeline. So I think it makes a ton of sense that you meet users where they are there. And it's just one of many, like, chat platforms we support or plan to support. It's just the one that is like...

1:08:06has almost 100% overlap with our ICP today. Yeah. I hope it doesn't get shut off. Just in terms of building with Slack is fun, right? Yeah. For tools like yours. Well, I think their documentation could do some work. And the app approval process is quite painful. But they are rolling out some pretty cool and interesting new features to support agents. It's much easier to stream messages.

1:08:36to Slack or stream status updates or tool calls into Slack response than it would have been previously. I think the previous mechanism would have just required repeated updates to the same message. So I think we're still, I actually am not seeing those features widely used right now, so we're in the early days probably. But I'm sure Slack is not going away, probably mostly because of the Slack Connect feature.

1:09:00the network that it creates between organizations. Definitely. To get critical, like, what's the... I always like to ask about pre-mortems, right? So if Sahzabi were not to, you know, IPO become the... You know, if it were not to become the standout in our, you know, existing fund and many others, why not?

1:09:26What's the story you'd write today to predict that? And obviously, this can touch market, it can touch it. I can think of a few causes of death. One of them is maybe not predicting the rate of model improvement correctly. Either that stalls or it advances at a rate faster than we had planned. And then I think if models don't get better today, then there's probably still a lot of...

1:09:55you know, post-training and RL that we can do, evals that we can create. I think our product is mostly still viable, but if models, let's say models get twice as good or five times as good, like that will probably, I can imagine that might invalidate some of what we've built or even like the way just the software development, it would transform the software development life cycle. So much so. So much so that like.

1:10:19it's unrecognizable. And that would mean that the type of observability that you all, like observability as a whole wouldn't be as necessary or like... There's one future I can think of or two futures that I can think of. One is the world of just-in-time software, where all of the software you've heard me talk about before, all of the software that is created is created on an ad hoc basis for a particular Session used and then thrown away and the model generates that software That could be either code or you know in a more even more futuristic and bizarre Version would be the generative UIs that are where the pixels themselves are being generated which we saw in thropic do a It proved a concept for almost like a year ago. Yeah, that's a world where I think observability doesn't make as much sense because you're not maintaining right software long

1:11:12or persistent software. Maybe observability takes on a new meaning where it's a focus on observing models and agents. Another feature is one where there's just zero software and interfaces go away entirely. And everything we do, we do through an agent. And I think this one is less likely because humans still love to see and click things and track things. And some, even as we use chat bots, it's useful to have a button that says, except edits.

1:11:41Those are those are two worlds where maybe observability as we know it like it's completely obsolete that I think much more likely is a world of like abundant software or infinite software where the amount of production software that is written and deployed and runs in perpetuity just continues to grow at an exponential rate or greater than linear rate. And the only way we can effectively monitor it is with AI, or agentic observability, platforms like Cezabi. So that's one potential risk. I think...

1:12:19There there's competitive risk a competitive risk like obviously other companies are focused on automating software development lifecycle They're generally focused on different parts if Robbock has cloud code, but I think mostly they're focused on model development Cursor has cursor and composer to and they I think increasingly do more model development, but they also live in like the middle stage of the software development lifecycle now. They purchase graphite, they have bug bot, they have peer review bots, which is one step before deploy, and then after deploy comes observability. So I think there are other companies that are, fortunately, they have other focuses and more competencies. So I think you'll want a solution that

1:13:07The company that will win in the space will be explicitly focused on observability and on detection and root cause analysis Yeah, also like just building great storage for for logs like multi-region high availability high durability low latency storage You guys are already thinking about that which we are we already have It is like a hard problem And making that cost effective is a hard problem. So I think it's just like a big it would be a big Scope for one of those teams to take on and do well But that that that keeps me up at night obviously like the market turning I think it was weird or we we live in a very funny times with geopolitical risk and the price of oil and the war in Iran and the war in Ukraine It's unclear like where the market will be and six three months let alone six or 12 months

1:14:03There's I guess a potential for you know, just a venture market winter or an AI winter that could be damaging Maybe the cost of inference goes up to a point where the economics of this solution don't make sense Obviously Energy is becoming more expensive and scarce. So that's something I think it would potentially make something like so zombie on non viable I think all things considered like there are a few yeah there's not that much I can do about the meta venture risk. And I feel pretty good about AI. It's getting better at a rapid rate and has been for a while. I don't think that there are many other teams that are as well positioned to do this that have our particular product vision and our singular focus. So I would say that our odds are pretty good. Yeah, so much of my conviction here, you know,

1:14:59was about you, right? And you and everything you're bringing to this. but also the tailwinds are so clear, right, particularly in terms of the models, particularly in terms of the benefits that you have as an AI, like building this company now versus building this company two, three years ago, right? There's no ship to turn, right? You can build and I think adapt a lot faster than other folks, so it's very exciting. Yeah, I mean, everything that we wrote about in the memo, like everything I, when I go, I don't look at this often, but when I occasionally have to, I'm struck by how right it is.

1:15:32And I think we've actually opened up on this on this note, which is like the Just the things that weren't were counterintuitive or non-obvious Six or eight or nine months ago are becoming more obvious which is very valid if it means that we got started at the right time Mm-hmm. Mm-hmm. Absolutely And we see it all the time that companies are too early, right? You can you're so you can be right, but too early which means you're wrong You know everyone makes those Mistakes because you know ventures. It's a craft at best I think and you know predicting the future is what we all try to do It feels better when The stars align of course yeah, it's we were a little early on healthcare voice. Yeah With our previous company and so I know what it's like to miss the boat and then we were late on other things there were healthcare regulatory opportunities I feel like you missed so one thing that helps with predicting the future is

1:16:31being a subject matter expert in your space. Totally, yeah. Great piece of advice from her. Yeah, exactly. But I think if you look at Paul Graham's essays from years ago, probably over 10 years, about how do I come up with a venture idea. In some ways, it's like, what stuck with me from these pieces that are many years old? Okay, schlep work, right? Just schlep blindness. Yeah, exactly. Do something dirty and unsexy. I mean, it's actually one of my favorite essays in venture.

1:17:00And it's a little blind to exactly what you're able to lean into, which is the domain expertise. I think, especially today, as we see what's required to build, what we need has just the ceiling, I always forget. It's a lot easier to build than it used to be. It makes my metaphors. And because it's so much easier, in many ways, I do see and I hope for, we're seeing early signs of it, some renaissance of I don't know if that's a proper word, but you can imagine all sorts of domain experts are the ones to win today, right? They should be so much better suited. Speaking of which, your dad, your dad is an orthopedic surgeon, is that correct? Yeah, I'm retired now. He's been retired for about a year. Okay, so he's no longer doing surgeries, is that what that means? He's no longer doing surgeries or office visits, clinics. Okay. So he's out.

1:17:57It took a long time for us to get him there. Because he's a workaholic and loves what he does. And orthopedic surgeons, as you know, I've had run-ins with them. I got hit by a car, broke my leg. These guys love doing surgeries. Fixing bones. I've heard rumors that he's, or maybe I've seen this on Twitter, that he's getting into vibe coding. Tell me what's going on here. He's on his own.

1:18:26IT, programming, and computer science journey. Well, one thing he did when I was, I guess probably as far back as college or high school, was he discouraged me from going into medicine. Which is really interesting. What was that about? I don't want to, I can only speculate about his psychology, but I think my dad would have been a great business person. But he comes from a family of lawyers and doctors from East Tennessee.

1:18:56Maybe had certain set of like value conservative values built into his I mean conservative not in the political sense risk taking risk taking the appetite of risk and what like is a what's a respectable or responsible career or adulthood like I think he regrets some of that decision because he's I mean he's at the top of his or was at the top of his field and has been very high performing and I think it's just had to work very hard for a long time. And I've seen other people that maybe did not work as hard as he did do much better than he has done. And people who are interested in similar things, he was buying computer magazines when that was the thing. He was doing homebrew kits. Really? Yeah. He's always had a technical bent. And now he's buying Mac Minis? Or what's it look like? That's so funny.

1:19:54He's sending me open-claw videos and telling me like sure what I'm gonna buy you a Mac many for your birthday And I'm like dad. I've already got a Mac many. Yeah, come on. Do you think you think I'm an amateur? Yeah, I guess it started with I think his curiosity was piqued when I Was you know I was into online video games as a kid and I asked if I could build my own gaming PC. And so now I still have a Nvidia GeForce 760 box in my bedroom. Amazing. How much did it cost? Do you remember? It was probably, I'm going to guess it was $400. It was pretty big. It was by far the most expensive part of the rig. I think he could see that there was something positive there. He kept you off the streets. He encouraged me off the streets. I was addicted to something else called a World of Workout. Another kind of drug.

1:20:43So he you know to his credit encouraged it and then encouraged me to go to try this web development bootcamp when I could have But it wasn't clear like there would be a real payoff. Yeah, and then as my engineering career got started I think he found that very interesting and was excited for me and wanted to be able to talk to me about the type of work that I was doing or Understand what I was building and you know And that led him to take like some online computer courses at Awake Tech, which is the like local technical college. Is that Wake Forest? Nope, it's not Wake Forest. I think it's just Wake County Technical College, but Wake Forest is nearby. And now in retirement, he's had a lot more time on his hands and has been able to go a lot deeper. And he's probably done four or five courses at this point. I don't want to, he'll be upset with me if I undersell.

1:21:37because he's worked very hard on them, and built some impressive applications. He's built full stock applications, deployed them into the cloud. A lot of the technologies that they're teaching are technologies that are a little bit hard to use so that you would probably not reach for today, like VMs. So he really knows his stuff. He's taken AI courses on traditional AI with CNNs.

1:22:04He's learning AI concepts that actually even predate me. So it's fun to sort of trade notes. Then more recently, as agents have become popular, I've taken the opportunity to kind of show him how it can use things like cloud code.

1:22:25to redo all of the things that he's learned in the past over the past year or two. And at first he was, I think, intensely frustrated because it's like, what the heck? I just spent two hours debugging my pie charm ID. And it took forever just to get this simple app working and Cloud Code is just one shot at it. But then I think he's now at the point where it feels like a creative unlock for him and is able to build so much more.

1:22:55It's creates this amazing like learning a rewarding feedback loop that he's really running with and yeah, so he sends me projects that he works on it's my sister Projects we're in a group check text. There's my younger sister as an engineer as well. Yeah, where he will it's mostly him sending things to us Did he ever get onto the Mac mini? Is he on VMs now? I think he's got some stuff running in the cloud. I'm not sure if he's bought himself a Mac mini for a long time. He was taking his computer science courses on a Lenovo ThinkPad, and I was like, dad, we've got to get you out of PowerShell. You're going to just wait until you get a MacBook. Trust me. That's awesome.

1:23:35It's been really fun. Recently, his passion has been open claw. He's really excited about it, and I think he's been amazed at what the incredible use cases and how people are building and running entire businesses using open claw. It does make you wonder how his... Again, I don't know much about...

1:23:57His career practicing medicine, but but what would have changed right if if if he had access to open claw, right? Yeah, that's that would be interesting. I mean yeah orthopedic surgery in particular is so manual Although my dad has like an inventor streak. He's invented multiple medical devices Over the course of his career, which like if you spend 40 years yeah ACL repairs like you Wow, you come up with some good ideas, okay? I'm sure that AI would have made the process of designing and developing and and bringing those to the market. Yeah, 3D printers. On the subject to continue down the family tree, I guess. So your sister works with you, which is really cool. I've met her. She's rad. She's going to HBS in a couple of months. And in the meantime, she's getting really a front row seat to building in Silicon Valley, which is cool. What is it like working with her? How does she...

1:24:54Change you are you intentional about like you know what you want to impart to her out here? Yeah, I mean her so my sister name is Hadley. She's five years my junior Which I think is a really good gap because five years is like you're close enough that you you're friends, you empathize with each other, you have sort of shared experiences, but not so close that you compete. And Hadley, to her credit, I think has been a, you know, has maybe five years of professional software engineering experience under her belt at some top companies. Which means Brex as well. She was at Brex, but although we did not overlap, we missed each other by just a couple of months. And then she went to Doppel, which is Andres and Bax, cybersecurity company. And more recently has been thinking about transitioning into business.

1:25:40roles or operations roles, wanted to sort of explore what's available to her outside of engineering. And that led her to apply to business school. She's going to HBS in the fall.

1:25:50She'll probably still be involved with Sazabi while she does that. We'll see how long she lasts before she... I want her to have some of that experience, but it'll be a big loss for us. After getting into business school, she decided to come join Sazabi as their first non-technical hire, ironically, because she is an engineer. But she's been basically shadowing me and serving as chief of staff. In a lot of ways, because I'm a solo founder, I think she serves in kind of a co-founder capacity or supporting co-founder capacity.

1:26:18She's there to kind of jam with me on fundraising strategy, on product questions like pricing, or how to sequence our launches, or manage relationships with our PR people, or throw our coming out of stealth party, which was a massive success. Big hit. So she kind of fills in a lot of different gaps. I think she's honestly perfect for this role, in part because she has a different personality than me.

1:26:48She's careful and thoughtful and considered. And I tend to be a little bit more bold, creative, and excited. And I think there's a nice dynamic there. Also I think, obviously because she's my sister and because we have 26 years of shared history, there's a huge amount of trust.

1:27:13There so I can be really honest with her about Like challenges that I am experiencing or ways that I think her performance could be better or risks to the business I can be completely open with her and honest with her and I know that we both have each other's best interests at heart and It's never None of the feedback we share is ever critical. We're focused on making this as obvious succeed and helping each other grow. And so I don't know what to say. It's going to be a big loss. It's been amazing working with her. I've still got her for three or four months and then in a full-time capacity. And she's going to help me kind of backfill some other non-technical roles, which will.

1:27:59I think take over some of the responsibilities that she's been owning. But I'm hopeful that she'll continue to own with us and maybe join us again this spring or summer. Out of curiosity, did you approach her? Did she approach you? How did that approach her? I approached her. And this is probably, there's a pattern here where I'm...

1:28:19I'm often thinking about what might be a great next move for Adley, and I want to be mindful of the fact that that can be constructive, and I'm sure she appreciates that in some cases. It can also probably be super annoying to have someone who's overbearing or cultivating you.

1:28:37And so I always want to give Hadley opportunities and information and choices, and then to get out of her way. In this case, I told her that I thought... One of the best ways for her to learn about startups and entrepreneurship and operations and like how to launch something would be to come do it with me at Sizzabi, which just so happened to be at an inflection point where we needed someone with her with her skills. What's ironic is she's been going back. She's gone back to HBS once for like a, you know, admission, admitted students, visitor day. And they're like having lectures about how to start companies.

1:29:14Like what entrepreneurship is about and how these like I just like raised the seed round for this thing and like I launched the company So I think she's really it's it's funny for her to experience both sides of that Yeah, but I approached her with the with the idea and there was some hesitation. I think at first from both my parents and my sister. But everyone seems extremely happy with the outcome. Hadley has obviously performed incredibly well. And I think she's learned a lot. And she's loved San Francisco. This is her first time living here. So yeah, this was turned out to be great design. That's great. Yeah, we don't often see it. So it's cool to meet her and hear from both sides. How'd you hear about?

1:29:59Village why'd you choose village? You obviously had a ton of options So I guess when I kicked off the seed fundraise and that this video will come out when the seed is public So everyone will know the details of the raise and who was involved. We initially Erase some money from J2 ventures, which is a fund based out of Boston. They're best known for defense tech tech. I had a relationship with the team at J2 through Christine Kang who was basically an advisor to me during my first company. Not an investor. I think she was patiently waiting for me to find the right thing to work on. And then when she found, when she heard that I was working on Sizzabi, she realized this was it. I loved J2 immediately. I mean, I love Christine and Alex and John. They were very supportive from the get-go, even before I had a 40-page memo.

1:30:53The recognized good founder market fit I think and in a and a point of view and wanted to be involved My concern was that they were based out of Boston and that as a company that's you know being based in San Francisco and being in the DevTools market It was important for me to have a reputable reputable brand based here in San Francisco a team on the ground In San Francisco and so I was looking for a j2 and I really were looking for the right partner to bring in. J2's team knew the village team because they had done some deals together and put them forward as a potential option. And I was open-minded at a great conversation with Ann and with Ben, but in particular, I think it was my conversation with you that sold me on it. To this day, probably one of the best pitches I've ever done. It's the pitches pitching itself because you're like, you get bought in and then at some point we're just jamming on it.

1:31:48and riffing. And I think it, you know, it meant a lot to me that to have a partner who really understood my vision and knew why I was the right person to be building this company. And it felt like you really got me. Yeah, it was mutual. It was a, I remember the pitch. It was very easy.

1:32:09I think on both sides. Like you said, I understood the language that you're speaking, this sort of proto-thesis that I shared earlier around my language was headless BI. It's so funny. Often you're like, we're pitching the same idea, but we call them different things. And then we realize that's what's happening. It makes the pitch super easy.

1:32:33Like having pitched hundreds of times now over five years and sometimes a sales context sometimes in a fundraising context like you know that Like when someone gets it in like the first five minutes like it's gonna be like we're we're doing this. Yeah Yeah, exactly go through the motions, but that's how I felt with with you And I got to know the rest of the village team. Yeah, they're obviously exceptional It's they have a big portfolio of companies that we can sell into absolutely in all different spaces and a very strong pedigree Yeah, and I like the village model. I think that's what we're trying to do with angels. They're trying to bring in lots of people and create like a supporting cast of characters to help make makes this obvious a success. Well, thank you for that. We're thrilled to be working with you. And I've said this to you personally, but one of my favorite things about working with you is you have a plan.

1:33:25and you execute on the plan, like according to schedule, which is so hard to predict, right? Again, this comes back to, I think, the extent to which you lived and breathed this problem for years, thought really deeply about it and then...

1:33:40There's so many pieces that are required to build a successful startup to hit PMF, and you're on the path there. But you bring so much intention to it. It's a surprise, right? Investing is a surprise, because you don't spend too much time with folks, right? You make a snap decision, and then you have to spend years getting to know each other afterwards. Yeah, and sometimes we're just gathering data points the whole time.

1:34:09getting, you know, hitting the head a lot as a, you know, playing lacrosse, playing ice hockey. You seem to learn quickly. You know, you have a growth mindset, which makes it a joy to be a part of. So, this is great. I'm so excited to be on board. This is just the beginning, right?

1:34:34Hey, this is Ben Keznoka, co-founder of Village Global. Thanks so much for tuning in to the Village Global podcast, where we go deep on all of the biggest topics in tech. If you enjoyed this conversation, please subscribe to our YouTube channel. You can check us out on Spotify, Apple, wherever you get your podcasts. We'd love to see you for the next one.