From Coding Machines to Leading Humans ft. Leonid Igolnik | Ep. 21

Viktor Gamov talks to Leonid Igolnik (Former CTO at Clari) about his career in B2B SaaS engineering leadership. Leonid’s first job: teaching kids Pascal. His challenge: changing buyer behavior and scale complex systems.

Books mentioned:
► Influence without Authority: https://www.amazon.com/Influence-Without-Authority-Allan-Cohen/dp/0471463302
► Drive: https://www.danpink.com/books/drive/
► Blink: The Power of Thinking Without Thinking: https://www.amazon.com/Blink-Power-Thinking-Without/dp/0316172324

SEASON 2
Hosted by Tim Berglund, Adi Polak and Viktor Gamov
Produced and Edited by Noelle Gallagher, Peter Furia and Nurie Mohamed
Music by Coastal Kites 
Artwork by Phil Vo 

•  🎧 Subscribe to Confluent Developer wherever you listen to podcasts. 
• ▶️ Subscribe on YouTube, and hit the 🔔 to catch new episodes.
• 👍 If you enjoyed this, please leave us a rating. 
• 🎧 Confluent also has a podcast for tech leaders: "Life Is But A Stream" hosted by our friend, Joseph Morais.

Transcript

SPEAKER_02 0:00

Convincing your customer it will be better for them to use this is an incredibly challenging but also incredibly rewarding journey. But some of the trends we're talking about are very consistent and can be used to go pick your next opportunity for yourself. The hardest things we're all get to do as leaders is try to change human behavior.

SPEAKER_01 0:21

Hello and welcome for this episode of Confluent Developer. I'm Victor Gamov and I will be your host for today's exciting episode. Today's guest is Leonid Igolnik. Leonid is an incredible person to have a conversation with. I consider him a friend, mentor, and all around great person to have you in the corner.

SPEAKER_02 0:48

It's good to see you again, Vic. It's been a while and uh we don't connect often enough lately.

SPEAKER_01 0:54

Yeah, but uh we always every time when we connect, it's still um uh always good time, always great time. Leonid, so in this podcast, in the conflict developer podcast, we're trying to um find it's a little bit of research of origin of the modern luminaries or the modern uh leaders who establish industries and we like to learn a little bit more, not uh as a kind of like what they do right now or what's your take on AI. Obviously, we're gonna take a talk about this because it's a hard topic. But we start this conversation usually to understand where are people coming from. So, and our usual first question to start this um to start this recording, to start the discussion is first of all, um, in your words, uh what are you like who are you and what are you doing? And after that, we're gonna go into this uh next set of questions.

SPEAKER_02 1:48

I guess professionally I consider myself an experienced engineering leader in a B2B enterprise software space. Uh, that's the space I'm uh been playing in. Actually, interestingly enough, two weeks ago, it's exactly uh I celebrated 20 years of leading software teams at various software companies. So I I guess now I can officially say I consider myself a professional in the uh uh engineering team uh leader in a B2B SaaS space.

SPEAKER_01 2:18

So how how you said the 20 years, it's a lot, it's a lot, it's uh it's a huge, uh, huge number. How everything started? But let's start not with your first job in in IT, but let's start with your first job that you you know earned some money. Something that you've done uh and people loved your services, so they will be grateful and they paid you some uh dollars, rubles, shekels. I don't know. We have a different uh uh different currency in this podcast already.

SPEAKER_02 2:47

Well, I don't know if they loved what I done. My real, real first job when we just immigrated to Israel uh was stuffing people's mailboxes with uh advertisement for Chinese uh takeout restaurant. Again, uh while I got paid for that and I learned the the the city of Haifa like the back of my hand. I don't know if a lot of people that whose mailboxes got that appreciated that. But my first kind of more full-time job, interestingly enough, was uh a few years later when I started teaching kids at the local community center. I started teaching Pascal of all things. I'm old enough to know that language.

SPEAKER_01 3:25

How you end up learning Pascal to the level that you were comfortable of teaching other kids? How it started?

SPEAKER_02 3:32

I discovered computers when I was probably 12. And uh, like a lot of people in our profession, I'm a trained extrovert, and I discovered that if I learned this programming language thing, I don't have to convince humans to do something for me. I can go write something to the computer and it does it, right? That's incredibly rewarding. And I started with basic, and then very quickly at the local uh either community college or something, they they had uh uh an after hours uh thing, and I started learning Pascal. I think I still remember writing uh Game of Life as the first uh thing, or other than boring stuff like sorts. And I knew Pascal uh for a number of years by the time I got to the point, I think it was uh first year of university where I started. I was comfortable enough teaching it to like 10, 12 year olds.

SPEAKER_01 4:24

Oh wow, that's that's really cool. So, what was the uh we were talking about like teaching different languages? I don't know if this the podcast that I spoke with, Bark would go first, or this one's gonna be first. But we were talking about different abstractions and how some of the abstractions that they teach in computer school are actually leaking in terms of like okay, so how you describe the object, is it horse, is it object, uh, or is it uh is it class? Um, and is it that horse is object or it's uh it's a class, or if the horse um, you know what was the the challenge as how you would explain the it's easier probably for Pascal because Pascal was the kind of like a procedural um language rather than object.

SPEAKER_02 5:05

In a way, I think the challenge was kids is not even about the concept of the language or teaching them the fundamentals, but finding examples that make learning to program a computer exciting, right? Like I still remember that first moment where I typed something into the computer and then you see the result and you go, wait, if I do this, it's gonna do that. It's getting that moment of discovery, I guess, in a way. Uh, that would have been the challenge. And quite frankly, I think it's the challenge in my professional career today to go full circle. You know, and my role, my job is to develop leaders.

SPEAKER_01 5:40

Yeah.

SPEAKER_02 5:40

And uh, what do we do? How do we find leaders in this industry? We take professional software engineers who are good at what they do, and we're gonna tell them, hey, listen, we're gonna take you from programming deterministic, fixed, variable, fixed feedback time things to programming very uh non-deterministic variable feedback time things known as humans. And by the way, you're gonna suck it at day one. So, how do we convince you that you actually want the job after you spent, I don't know, a decade, maybe more before you got into management of getting good at programming the other things? So I think the challenge is the same. How do you find that balance between the frustration that inevitably comes with learning of anything challenging, uh, but with the joy uh of that learning as well? And I had gone through that uh in my 40s actually. I always wanted to fly. I always enjoyed aviation, and uh I pursued my passion for flying uh in my early 40s, and I remember how humbling it is and how challenging it is to go learn a new skill, wealth of knowledge, to get a pilot certificate, and you know that was a good vivid reminder for me for what my early and career leaders are going through as well. So, how do you balance the frustration that inevitably comes with learning with the morsels of reward that you get during the learning curve itself?

SPEAKER_01 7:07

Yeah, I think it's also very much depends on uh on age and experience, because um sometimes we like to say that oh, I I need to have motivation to do the things. When you're getting older and you have a more mature professionally and you know the the physically, you uh you trade discipline with uh with motivation. So you don't need to be motivation anymore, you kind of a discipline because you know that you know there's a resource, time is limited resource, and how you're gonna be using this as your kind of investment chip uh that where you will be putting this is very important. And this is very this is a very cool thing that you to you talk about this. Since you mentioned that you were um uh building the leaders, you're building leaders for different businesses. I I assume I presume because that's that's a whole point. We uh we we starting some companies or we're joining companies to build something. So let let me let me shift our gears into the main uh topic of this conversation is that we like to ask our guests is the what was the most uh kind of like a challenging thing that you ever done, or most uh difficult thing you've done, or you failed to do. Um and maybe you can talk about the your experience in you know human side and human NIT side as well.

SPEAKER_02 8:26

I think there are two dimensions we can go explore. One is uh the human side. Um the hardest things we all get to do as leaders is try to change human behavior. That's the reality of it. The goal of an engineering leader or any leader is to get this group of humans that have their own desires, needs, problems, pains, aspirations. And if you combine the vector of all of the directional movement at work, you need that vector to roughly point into the same direction, right? You will never get everybody 100% aligned, but like the sum of the sum of all vectors has to point into the direction you need them to go. So that's challenge number one. I think challenge number two is figuring out what's the what's that vector as an engineering leader, what's that vector we need to go point to to support the business. And for me in business, and I've been lucky to be at the right place at the right time several times and then several times deliberately, is changing the behavior of the buyer. Like I I really love creating new things, and I've been lucky enough to be in several uh industries where we've created new either new spaces uh or new classes of enterprise software or new spaces, and uh convincing your customers that uh it will be better for them to use this uh is an incredibly challenging but also incredibly rewarding journey.

SPEAKER_01 9:55

Give us uh give our listeners an example. I know like the places where you have been and the things where you've done there. For those of you, for those of listeners who don't know, like just give a few few, say let's break it down by kind of like a decades, like there was a decade of something, like when the like uh the people didn't talk about cloud, or even before, even before, or what's happening.

SPEAKER_02 10:15

I am old enough that uh you know some of my early experiments on in the cloud were called ASP application service provider uh or on demand. Uh so even before the the modern um SaaS uh world or the term. I think my first exposure for behavior change and how easy it can be in the right circumstances when when I joined a company called Tukav's OpenSRS. Uh, this would have been um early 2000, right around the time uh you were able to register your own domains through more than one company. Like back in the day, when I was younger, uh Network Solutions was the only company where you could go to the go and get your.com.org.net domain. And then the namespace got deregulated, and we were building products to help people sell the domains wholesale. And we had a network of customers and they were pulling us into that. And I saw us going that took us from uh selling domains to selling SSL certificates to sell email, selling email, DNS, etc. services. And the lesson I later extracted out of that is sometimes there's a wave that is pushing you into the market, and if you write that wave successfully, you can build a successful business. We were at the time, and roughly uh around the same time, GoDaddy as a domain registrar showed up, uh, who uh outpaced us eventually, but we were second largest domain registrar in the world uh at the time, and I watched without probably understanding I was in my early 20s, uh kind of a wave pushing us into the market. My next interesting gig uh was when we actually convincing the world uh that SaaS, as a software delivery model for industry, uh for enterprise software will be good for the industry. I was at the company called Teleo that eventually was acquired by Oracle. Uh 2007 to 2012, Teleo, one of the four pure play SAS, largest pure play SaaS companies uh out there in the market, it was us, um, Salesforce, uh, that started in that space fairly early. Another company in HR space called Success Factors, and another company called Concur in Expense Management. So this were we used to be called the four horsemen of SaaS. And we were going through a very hard journey of convincing customers that we can handle information security and protecting their data better, quite frankly, than some of our customers could. And we probably all know how the story ends. We all have embraced cloud and the SaaS, but they were conversations that we were still having. It was like, hey, you think I'm gonna put the salary of my CEO into your cloud system?

SPEAKER_01 12:57

No way, right? And uh, well, well, well, how the uh turntable, like uh to quote the famous manager uh from office. Um and so that was a time of the internet infrastructure, um, pioneering SaaS.

SPEAKER_00 13:13

Now a quick word from our sponsor. Confluent Developer the Podcast is brought to you by Confluent Developer the website, which has everything you need as a developer of data streaming systems. And it's completely free. We've got curriculum, hands-on exercises, executable tutorials, the online data streaming engineer certification, also free, a way to find a meetup near you, those are free. Everything is there. I really want you to be successful in your journey as a data streaming engineer, and this is the site that has what you need. Check it out at developer.confluent.io. That's developer.confluent.io. Now back to the show.

SPEAKER_01 13:53

So the next step was okay, so how we can observe the things, right? How we can get the metrics. And everyone is already in the cloud. How can we help them to run this successfully?

SPEAKER_02 14:03

So the most interesting journey I had right after that is joining a company called SignalFX, which eventually got acquired by Splunk, um, to lead engineering in a completely different space. And maybe the trend that the viewers should notice this I do jump markets a lot, but some of the trends we're talking about are very consistent and can be used to go pick your next opportunity for yourself. Uh, Victor, if you remember, you and I had this conversation about your choices one day, and we can double-click on that. But um, if you remember the first two, and I was, you know, the headline from the first two, trend is your friend. If you're riding some kind of wave, you can build something big. Uh, at some point the cloud became the norm, but the observability tools uh that were built, um, the the wileys, the dyno traces, the dynamics of the world, were not sufficient uh to observe more modern, more ephemeral infrastructure. So Docker adoption of containers created much more uh hard-to-observe sets of environments. They're much more dynamic than environments that used to you know change once a quarter when you do a software release. Yep. So we had a wave and we had a need of this new observability tools that was uh pushing us into the market. So uh at the time, the leaders were wavefront, signal effects, datadog, light step, honeycomb, and maybe the other lesson I've extracted for this is much like with Staleon success factors in the HR software race, we ended up being number one and number two. Uh, you want to make your bad ideally on something who can become a number one and number two, or fundamentally set up to be number one or number two. And that's what happened with uh uh Datadog and SignalFX. Datadog stayed independent, we got acquired for about a billion dollars by um Splunk. Uh also another trend that was pushing us into the market, and we were riding away and riding away in a completely different way.

SPEAKER_01 16:00

At SignalFX, the challenges were not convincing the market, but scaling our software architecture that probably it's kind of like opposite because uh in the first uh the first time you need to convince people, but people already were convinced. They just wanted to, you know, use more and more of the software, are like bringing more workloads to be observed through the system. So scaling this.

SPEAKER_02 16:21

That's right. Like the first gig I told you at two at two guys, we were being pulled into the market at Toledo, we were pushing ourselves into the market. Signal FX was being in part pulled into the market. Um, and then uh about three years ago, coming up in three years, I joined another company where I saw another waveforming. Uh, we at Clarity we built tools for sales professionals to run their processes better. We've been leveraging AI before AI was cool.

SPEAKER_01 16:46

And uh speaking about like a putting uh putting your CEO salary in your system, you're not putting that not anymore, CEO salary, you're putting whole information about your company and the businesses and how the people, you know, how the sales motions are working and uh how much money company does.

SPEAKER_02 17:04

So we're sitting on a lot of the sensitive information uh for a lot of customers whose name your viewers and listeners will recognize. And uh this is an example of creating a new market. If you look at the sales tooling market, the space has been somewhat stable for a while. CRMs, the do major CRMs, HopSpot and Salesforce have been out there for a couple decades. And again, much like when we were building stuff at Teleo, HRIS systems existed before us. We just built a better system of insight on top of system of record. In a way, that's what we're doing at Teleo. We at Clary. We're also building a system of insight on top of a lot of system of record, applying at the time at Teleo, we were applying cloud to the problem. Now we're applying uh modern uh machine learning and AI techniques to the problem. And uh it's another opportunity to push, and we're very proud to say that a year ago this market got recognized by one of the key industry analysts, Forrester. In a few weeks, uh Gartner is also going to publish something on this space as well. So, this is a journey of convincing the market that those tools are good at you, and again, if if you see a pattern, if you can become a number one or number two, uh the rewards are lucrative for the space builder.

SPEAKER_01 18:22

Apart from being kind of like a very um, I would say observant about industry and understanding the trend, what was your like a personal um um challenge to maybe adopting or maybe adopt uh minds of other people, helping them to grow because like building the teams in the highly efficient teams, uh building new generation of leaders on these different stages of the company, it's very important uh for the success of those companies. So I want to kind of like a focus a little bit on some of the things that you observed and how you kind of like manage to uh adapt.

SPEAKER_02 19:03

So it's interesting. I still consider myself a technologist at heart, so I can probably uh sit down and kick out with you and the listeners on how you train LAMS and how neural network works.

SPEAKER_01 19:14

Uh, although you're still you're still doing the conference presentation on the practical uh practical aspects in using some of the things with the intent integrity chain that you did with uh with our common friend and some other things.

SPEAKER_02 19:29

So so while I I I continue to keep up as a technologist, and I've been lucky enough to have a pretty good fundamental education that allows me to understand everything we do at the highest levels, all the way down to possibly maybe close to silicon. Um, what I've spent the last 20 years now that we passed October 5th of my career is reading more psychology books, believe it or not, than technology books. Because when it comes to programming or motivating humans, there's there was science to it and there's art. And uh it's all about uh creating an environment. And uh, Victor, you heard this book many, many times. You know where I'm gonna go with this. My favorite uh topic on this is uh from Daniel Pink. The book is called Drive. Uh, and it says, hey, if you want to create an environment where knowledge workers thrive, and by the way, that applies not just to leaders, it applies to individual contributors as well. You need to create an autonomy of autonomy, mastery, and purpose. And my new my new variation on this is you have to amp it up AMP autonomy, mastery, purpose. Um, in order to do that, you have to influence, and sometimes you have to influence with authority, i.e., you report to me and let me go nudge you, right? Into this. Uh, or you have to influence uh without authority, and again, I'll shameless plug for another book if you want to add to the show notes a Bradford Cohen uh book called Influence Without Authority, another book that teaches you a set of techniques of modifying behavior without having direct authority on the behavior. And lastly, you do this by being able to articulate and tell a compelling story. We as human beings have been wired to listen to stories and use our imagination of stories. And if you remember my foray and telling you how painful it was to learn how to fly, the most painful thing was to learn how to land an airplane. That's the the hardest thing we'll all get to learn how to do as pilots.

SPEAKER_01 21:30

And it's the most important thing, as a matter of fact, right? So it's not uh takeoff, it's kind of landing.

SPEAKER_02 21:35

Taking off is much easier than landing, but yes, if you want to complete the flight, you have to land, and everybody in their training typically reaches that plateau of frustration where, like, I can't get this plane to to do what I wanted to do when I want to land it. And that happens in professional development as well. So the question is how do you motivate the person to plow through the pain? For me, it was imagining why I'm doing this, right? I use the airplane to travel with friends, I have some friends visiting for. From out of country this weekend, we're gonna grab the plane, we're gonna go somewhere, right? That was my purpose. Remember autonomy mastery purpose?

SPEAKER_01 22:08

Yes.

SPEAKER_02 22:09

So I needed to achieve a certain level of mastery. And those are the techniques that I've used successfully with a number of uh folks that worked on my teams to push them and sometimes to learn, you have to go through some very painful experiences. Uh, and as long as you understand why you're doing this, what's at the end, right? Uh I think that the pain can be worth it. Um, and that's kind of my approach to understanding people that do, and you can only do it with people who are interested in that mastery. And to be fair, not everybody at every point of their professional career has the same ability to spend energy on mastery. We'll go through different things outside of work as well. But those are the techniques that I tend to leverage um when it comes to helping people grow professionally in that works with the earliest in-career fresh grad, all the way to people that had 30 years of experience.

SPEAKER_01 23:04

So you I think the if it's summarizing, the challenge right now for you is to stay uh technical enough but still um having scale uh teams by implementing certain certain things that you learn in life and how you can successfully and efficiently pass it to um in other generations. And you're absolutely right that when um different people will uh react differently, there's uh it could be uh pushback. Some people would like to stay where they are, they might not be um uh they say that I'm not gonna grow. And this is kind of like a wired with our human body. The people's brain likes to be, you know, left alone. Uh why it is very difficult for you to learn something is because you're forcing your brain to develop new neural paths in order to learn this skill. So that's why mastery takes so much time because it is it requires you time to train those skills over and over again. You cannot um you know this kind of like a famous joke that uh it's nothing to do with flights, it's just like a testing people's uh confidence. And the journalists who are asking people on the street how many of people without proper flight training, how many people would land the plane? And what how confident are you that you will be able to land the plane? And it is the same thing with the with the um I think it's very similar with um um with kind of like fake it until you make it type of approach, right? But but at the end of the day, there's a lot of things um get connected with this, right? Um so what's your um what's your advice? I don't know, like when you would the the mentor people who want to do develop this um taste for sporting new trends. How you're gonna sport the new trend in the in the industry? How you sport new trend, you've been in the different like stages, how you see, and I wanna you bring back a conversation that we have in May 2017, uh when we spoke about um at the time when I was like between the jobs and I was considering some of the other opportunities. So it's kind of like maybe it's a framework that you have, maybe it is something kind of like a gut feeling, you know, you're playing by E or something like that.

SPEAKER_02 25:42

I think for me it's a couple things. Some of it it's the framework that covers some of the patterns that I mentioned. Trends your friend. Uh ideally, you want to go build something that pushes you into the market rather than pushing yourself into the market. Both can be lucrative, but one is easier than the other. Uh, understanding the is the buyer ready to consume? Like, are they ready for the change? Because I said the humans don't like change, buyers don't like change things, right? If they if you like, you know, I've been pitched on a number of new observability companies. I'm like, well, we just took an industry through a change. Do you really think the buying public is ready to go re-re-architect and re-embrace uh uh the monitoring stack? And if they are, why? What would drive them? Is it cost, performance, or is it an ability to do something? So turn to your friend. Maybe that's headline number one. Number two, experience matters because um, at least my experience of 20 years in leadership plus some years as an IC demonstrates that there are a lot of repeatable patterns, and at some point, uh, I'll go back to another shameless plug of a book, Blink. I just rely on the innate patterns my brain has learned to recognize. Whether it's about setting up engineering teams and what works, what doesn't work, what good should look like. You know, uh, does your bill take an hour or two minutes? What's your code coverage? Uh can you push to can you hotfix? Can you roll back, right? Patterns of what good looks like. And then uh uh understanding the broader business trends that are going on. So I think we went through a decade of SaaS, or a couple decades of SaaS. I think the next few decades were uh decades of AI. That's one of the reasons I went back to uh to work to like a mid-sized startup, because if you want to stay current, you in like, and this is the most painful and exciting part about this industry, you have to learn new things constantly every every five years, the way we build software changes, right? And this new new new new white coding trend is no differently. We just we add we keep adding layers of abstraction and complexity and understanding how to navigate them. Um, and you know, we'll see whether the the projections or whether we're gonna need less humans or not uh to build software will be true or not, but we will need humans, and understanding how to pick the right humans to go on this journey with is also incredibly difficult. Ideally, you want to be the dumbest person in the room.

SPEAKER_01 28:22

Can you elaborate? Can you elaborate like why you want to be uh in the dumbest person in the room?

SPEAKER_02 28:27

Uh it goes back to I'm picking it up, eternity mastering purpose, the brain, at least some of our brains, and again, everybody has different energy balance uh what they spend uh at work and outside of work. Um, I strive when learning new things, right? Uh Victor, I know you spend a lot of time in the gym. You probably do it for the same reason. It's painful, right? Yes, it requires dedication, commitment. But yeah, what is it? You have to tear some muscle tissue to build new muscle tissue? Uh yeah, same exact same happens with reviring the brain.

SPEAKER_01 29:00

Same thing with reviring the brain, surround with the people where you can learn uh a lot of things, and maybe if you feel like you you hit this place where there's not much learned or not much to give, maybe it's uh there's something new available somewhere else. So that might be you know time to um to explore something new. And it's also I think it's also generic. Um what's the word I'm looking for? Like not a generational, it's kind of like the human evolution evolutionary, what word I'm looking for. So the evolutionary. So you kind of like you stuff when you start stagnate. This is the time where like you know, everything, you know, we need to put our our our hands, and we're just like, okay, we give up, we we're not doing anything. So, but before that, it always the learning always will be a struggle. Um, and yeah, so the trend with AI is is also very interesting. We spend a lot of time to explain uh buyers how what kind of value they will get with uh with AI, implementing those things. So definitely, um definitely the trend is here, it's a very interesting trend, and I'm actually looking forward to learn more about and how this uh will affect us and how it affects industry. We already see a lot of things. And I'm I'm excited to um to also see like what you are you you will be doing for the next you know 10 years uh and what the what the industry will be will be doing. Uh Leonid, any um any other kind of like a last last words, last words, um anything to to summarize or maybe you know leave uh leave our audience with uh a bit of advice or something like that?

SPEAKER_02 30:45

I think remember about the trends, but also remember about so-called Gartner hype curve. If you haven't seen that one, look it up, google it, understand where each of the trends you're following is on that hype curve because I've been in the valley for 20 plus years now. The hype curve is the thing, and it keeps repeating and repeating and repeating, and we're in the middle of another one, right? We're at the peak of excitement about AI, we're gonna hit what's called the thross of disillusionment, and at some point we're gonna climb back to the plateau of excitement, uh, plateau of enlightenment.

SPEAKER_01 31:17

Yeah.

SPEAKER_02 31:18

Um, so understand the hype curve, keep learning, uh, focus on autonomy, mastery, and purpose regardless. And if you if you're in the position to choose between jobs, ask yourself which one will satisfy those three. I think you're gonna enjoy it more. That's my headline. And specifically on AI, uh, I actually uh over this weekend saw an interview with Andre Carpathi, the guy that was at Stamford, got Tesla into autonomous driving. And he's not working for any AI lab, so he doesn't have any product to sell. Uh, it's a great two-hour interview with Andre talking about how he thinks the AI will evolve, what it's capable of, what it's not capable of. I do think the progress we're making and the inflection point uh we're making for the industry is meaningful, but it's also not end all be all that the labs or the vendors will want you to believe uh it is.

SPEAKER_01 32:19

Uh what do you mean? I I feel for me, it feels like again, being very uh technical and trying to spot uh waves and trends, it reminds me a lot of uh like a web search hype uh back in the day, like the in how the web search transformed how we're thinking about internet. You know, remember the time that you were thinking about internet as a domains, DNS. Now you're not thinking about internet in this matter. You know that you know using the search engine to find something. And the AI at this uh we still use AI as like a faster search or something that faster extract our facts and maybe generate something for us that we cannot do with the search. Um it's kind of like it's not creating anything new, uh a little bit, but it still requires some of the some of the no the the previous um data to to explore. What do you think?

SPEAKER_02 33:11

So so I think there are some inherent advantages of the current techniques, and again, Andre in his interview goes into a much more eloquent uh way of explaining this, but like one of his examples that struck me was like ask chat GPT jokes. Uh it only knows three styles of jokes, right? Uh, because effectively, if you think about what modern LLMs are, they are a compression algorithm for the text knowledge that was given to them during training. That's that's like right exactly. They don't know facts. We're building systems that allows them to equip them with facts. I think they provide a lot of value. I actually spent uh five hours yesterday planning uh a December holiday together with a couple of my friends and deep research chat GPT, and those tools were incredible in terms of making me more productive, it's giving me better tools. You still have to learn how to wield those tools. By the way, the same applies for the tools we give to software engineers. Um and it is absolutely a breakthrough. Like the search was a breakthrough, like Web2.0 was a breakthrough, like mobile was a breakthrough.

SPEAKER_01 34:18

The big big part was the the the the um user-generating content, and I think we're getting into this inflation point when it's like AI-generated content will be also prevail at some point. Hopefully not, because there would be um you know garbage in, garbage out type of situation.

SPEAKER_02 34:37

I think the industry is adopting the term of AI slop. There's a lot of AI slop out of there.

unknown 34:42

Yes.

SPEAKER_02 34:43

Okay, Sora is fun. I enjoyed the it's actually freaky how good Sora is.

SPEAKER_01 34:50

But also think about this. I've been uh if we get returning to your brain, and now your brain, it was exciting a couple months ago, and now your brain can spot on the spot, like reading something, visually looking at this. You can spot was something that was initially written by humans or something that was written uh using AI.

SPEAKER_02 35:09

It's getting better and better, like it does worry me, especially the video stuff. Like you can't believe everything. We used to be able to believe what you see, and uh even that is taken away from us. Yes, but but you know, every inflection point created a bubble: internet, cloud, yeah, mobile. Yes, uh but it also at the end, if you look at the history of our industry, it also left us with a very fundamental set of technologies to go build the next level of abstraction. Maybe the the place to finish is AI would not exist without cloud because that allowed us to aggregate all the content. Web 2.0 gave us the content. Uh uh mobile gave us a lot of visual content because everybody has a camera, and none of this would have existed if internet didn't exist. So the the points of inflection and iterations continue, and we're just busy building the next level of abstraction.

SPEAKER_01 36:07

Yes, this is great. This is a great one. Danit, thank you so much for being part of Confluent Developer. Uh, ladies and gentlemen, it was Lenita Goling from Clary. Uh Victor Gamo from Confluent. Stay tuned for a new and fresh episode of Confluent Developer YouTube channel. And as always, have a night later.