AI will reshape knowledge work. What leaders should ask, do and learn now

Eiso Kant, Poolside.AI: I think fear is the right word. I think I think fear is there as a mechanism to make us alert. And I think it's important, especially in AI, not to be yelling bomb in a theater every year about AGIs around the corner.

We are living still in a human-led, AI-assisted world, one where we have various degrees of capabilities in AI, from the ability to write code or language or use it to even, you know, write a song for your child.

But the reality of it is, is that the models are increasing so fast and capabilities that we are going to end up in a world that is no longer developer led and AI assisted, in a world where we have to reckon and ask ourselves the question when we have a AI that's more capable than us in doing the tasks and objectives we give it, what does our role become?

If we know that A.I. is going to become an increasingly larger part of our life, of our work life and of our personal life is to just spend time with it. Because as you're spending time with it, you're going to see the progress and you're going to be able to start answering, you know, parts of these questions over time in bits. As you see, that is becoming a tool. And it someday maybe some of the fears that we have today are very warranted. But you can be part of the conversation then on how you want the world to look like, both to your leaders in business and in government.

Linda Lacina, Meet The Leader: Welcome to Meet The Leader. I am excited to introduce you to Eiso Kant. He's the founder and CTO of poolside AI. He's going to talk to us about the advantages this technology has ahead. How are you Eiso?

Eiso Kant, poolside: I'm doing good.

Linda Lacina, Meet The Leader: Why don't we talk a little bit about, in broad strokes, the strategic potential of AI? I think that's going to set us up for some of our conversation here. There's a little bit of a gap between the strategic potential of AI and really how we're employing it right now for the most part. Can you talk a little bit about how we're kind of most people are using it now?

Eiso Kant, poolside: So, if we take your two questions from what is the strategic potential of AI and what's the gap that we see today and versus how is AI being deployed today? AI today, is frankly, in almost all manners, employed, employed as an assistant. Right. We are living still in a human-led AI assisted world, one where we have various degrees of capabilities in AI for various different types of activities, from the ability to write code or language or use it to even, you know, write a song for your child.

And so, what we're finding is that we are starting to understand what the capabilities currently are and we're deploying them where they're useful. But the reality is that across the industry, Poolside and several other players have kind of converged on the same point where we can see massive acceleration.

And I always kind of asterisk this with the fact that the last couple of years I haven't been running around saying AGI is 12 months away. But the reality is, is where we sit today from industry building foundational AI, is that we're looking at an 18-month optimistic, and I would say 36-months pessimistic, window in which AI is going to become as capable as, frankly, all knowledge work that humans today are doing across a whole host of fields, starting with the ability to write code, but also, frankly, in lots of other areas that we frankly today employ people.

Linda Lacina, Meet The Leader: As we're thinking about this, what are the ways that we should be using or could be using it that, you know, we're going to have to work up to? What does it look like?

Eiso Kant, poolside: So if we take this 18-to-36-month window, what we're going to see is a very rapid acceleration of progress in AI. We're going from places where AI might be slightly useful, delivering you 10 or 15% productivity, if that's in writing code where poolside is, you know, heavily focused oriented or if that's in helping you in your day-to-day basis in in your role as project manager or journalist or leader.

The fact of the matter is, is that everybody today, in my view, should already start using AI, because what is happening is that the progress curve is happening so fast that what AI might not be able to do today, you're frankly very likely going to be able to see you do in the next several years. And so starting to get a sense of where the capabilities are today for you in your role, have a conversation. These models and AI is currently in most cases, you know, deployed or at least the type of the AI we talk about in a way that you're able to have a natural conversation with it and start trying it out in different areas that are relevant to you, either in your personal life or that are economically valuable to what you're doing.

But frankly, I think the most important message that I've been I've been trying to drive home is that we have a really hard time of living on an exponential. We are currently in a moment where leaders like myself in industry are kind of standing up and saying, guys, us, our own timelines have just compressed massively and we are going to look at a world where intelligence is no longer something that is only our domain. It's something that we can scale up on compute.

And this is going to have a lot of second- and third-order effects, a lot of positive effect in terms of the amount of progress we can make in everything from science to driving down the cost of goods to its raw materials over time. But it's also, frankly, going to have negative externalities that come from something that is happening so fast. And so that's kind of the message I hope to try to get across here.

Linda Lacina, Meet The Leader: Why don't you tell us a little bit about what poolside, about poolside AI, what it does and the problems that it solves.

Eiso Kant, poolside: So poolside is a frontier AI company and we're fundamentally in the race towards AGI, closing the gap between human capabilities and AI and where the models are today.

Now we've decided to take the path towards AGI by focusing first on models’ capabilities to write code and build software and to reason. And, and frankly, our view has been since since early on that the first major economically valuable capability where we're going to see the gap close between human capabilities and AI is going to be software development and software engineering. And that has a lot of effect in the world because frankly, a lot of the world's automation and progress comes from software.

Linda Lacina, Meet The Leader And how does your technology work?

Eiso Kant, poolside So we are building foundation models. And so you might have heard of large language models, foundation models. And the way I often at a high level like to think about what is happening in these models that we're training to become more capable -- artificial intelligence -- is that we take extremely large orders of magnitude of data, far larger than it requires a human to learn anything, think the scale or size of the internet. And we're compressing it into what's a neural network.

And that compression of data on a neural network forces generalization, forces learning. And so to the end, this is how we build them. And there's a lot more to this. But to the end user, the way they work is that the interface to them is incredibly natural, right? Something we haven't seen before. We can actually talk in language with it. And the reason these models first became most capable in their ability to understand and interact with language is because that's the massive dataset that we had. The internet has been the first place that we've used to train these models, and the internet is full of massive amounts of language.

But what the internet didn't have was the thinking and reasoning change that we as humans create before we put the output work, right before we write the article that we publish. And it didn't have some of the feedback that we get in our own learning. And in the domain of coding, when you're learning how to code, you have access, of course, to resources that you're reading, but you're also actually doing the act of writing code. And when you're wrong, your computer's telling you when you're wrong. And it's that very deterministic feedback that allows us to make AI more capable.

And in our in our field, we pioneered the use of reinforcement learning from code execution feedback. The ability to take something like coding that is deterministic, that follows rules that we can simulate and create the world's largest simulated environment, over half a million real-world software projects where the AI goes in, it attempts different solutions and it learns when it's right and wrong. And what we're seeing just like a human, except that many orders of magnitude more skill data required, AI is very rapidly progressing its ability to think and to actually create the correct output.

Linda Lacina, Meet The Leader: And how will this change how a human is involved in writing code? Maybe even in a few years. How will that evolve?

Eiso Kant, poolside: That's a very good question. Today we live in developer-led, software developer-led AI-assisted world. And in that world, what we're seeing over the coming months and years is that the assistance becomes increasingly more capable. We're able to give it higher level objectives and have a lot more reliability in the answers. We're able to give it harder tasks where it can explore different solutions for us, often faster than, than we possibly could ourselves and get to correct outcomes.

Now right now, we're still at a state in the world where AI is incredibly useful, driving anywhere from 15 to 30% productivity gains. But the reality of it is, is that the models are increasing so fast in capabilities that we are going to end up in a world that is no longer developer-led and AI-assisted, in a world where we have to reckon and ask ourselves the question when we have a AI that's more capable than us in doing the tasks and objectives we give it, what does our role become?

And frankly, that's a long discussion, probably more than than, you know, this short podcast. And it's frankly the question I think that should be on a lot of our minds because it's not just up to AI leaders to figure this out. I think this is frankly for all of society and also, frankly, for government to be thinking about, because it will have both massive positive economic impact, but also negative externalities.

Linda Lacina, Meet The Leader: What are some impact stories that you have from poolside AI to kind of understand the sort of the efficiencies it has and also its potential?

Eiso Kant, poolside: So, so what we're seeing today is that we can come into enterprises. poolside today is exclusively available for enterprises that have anywhere from 3,000 developers up to 60 or 100 or even 200,000 developers in, in scale. And we deliver our entire solution inside the enterprise. And so, part of our promise for enterprises is that the AI is coming to your data. Their data is not coming to the AI.

And what we're seeing in practice is that we're putting this in the hands of developers. This is not just about a model we often refer to, right, as the artificial intelligence, but it's also about the end-user experience. And so, we provide an end user experience for software developers and their editors where they do their daily work in the web browser without other tools so that they're able to both do their own work much faster, but also to actually have a conversation with the AI, asking it to do tasks for them, asking it to explain new code bases that they might be in.

And what we're seeing is that currently the world of AI and productivity for software development sits anywhere from 10 to 30%. And if you think about software, you know, organizations, engineers are everywhere. Most people don't realize that most of the largest global banks have anywhere from 30 to 50,000 software developers per bank. And so this holds true in lots of industries. So places where we can start making accelerations that today are in the 15 to 30% range but in the coming years are going to become increasingly orders of magnitude larger. Frankly, you can have a lot of impact already today and see it at the end of the day in your, in your balance sheet.

Linda Lacina, Meet The Leader: And how will this change the labour force for those who are in software development? You know, how would it maybe change the number of people, right, will it shrink, but also the type of tasks that they'll be doing?

Eiso Kant, poolside So we take software development, but we can even take this broader for all knowledge work that's happening in the world. We're on the frontier, so we have to see how this plays out, but if the predictions that I'm making, our leaders in the AI are making, and we see that we've reached this point where AI becomes as capable as the most capable software developer, most capable lawyer, the most capable accountant, we have to reconcile the fact that we have something that we can spin up on compute, that runs 24/7 a day, doesn't get tired and is able to actually deliver that economic impact. And we have to ask ourselves at what does our role become in that?

The reality of it is, is that we are in a world where we can do so much still to create abundance. And so much exploration is still ahead of us. And so, I have no doubt that we will find a place and purpose for ourselves in that journey. And there is a lot of parts of today our role that actually is, is uniquely human. The asymmetry of information that we might have between each other, the fact that there are certain things that we frankly want to speak to someone face to face. But also, the reality of the matter is where in the next 10 to 20, 30 years, we might have an incredible robotic surgeon that we trust more over a human, we still want the nurse.

Linda Lacina, Meet The Leader: So, I think a lot of people listening to this or watching this will be asking maybe the question I'm going to ask you right now, which is, has coding sort of gone the way of the horse and buggy? How should we think about this?

Eiso Kant, poolside I think we should be thinking about, in very short order, all knowledge work. Everything we frankly do behind a computer screen is going to have to be rethought. And because we're going to have to reconcile the fact that we will have technology that's going to be as capable as us.

So, and I also don't think that we have all the answers yet, including myself. The world that we want to live in is a world where all of us have purpose, where all of us can feel valuable. But we also want to live in a world where we can make accessible to everyone the things that we can drive the cost down for, housing, education, health care. And you can go down the list.

So I'm very optimistic about what's going to happen. But at the same time, we do have to reconcile the fact that when in history, and we've seen this throughout industrial revolutions, there have been a lot of people left behind. And frankly, for this time is the first time in history where we have a technology that is going to impact so many of us at scale.

Linda Lacina, Meet The Leader: Our Future of Jobs Report that was just released [in January 2025] had a couple of interesting statistics. One is that 40% of skills are going to change. Another factoid from this is that economists are sort of saying, hey, you know, it might be more interesting in the years ahead if you're talking about skill and task growth over job growth, right? What do you think about that? What do you think about that shift?

Eiso Kant, poolside: With, with a lot of respect to, I think the work that's being done to try to figure out what the world looks next, I think it is still underestimating what it means to be able to. Today when we have a complex task, an objective, no matter if it's building a company, if it's figuring out a technology or trying to make progress in research and science, we put people on the problem because people are incredible, our intelligence is amazing and and we collaborate to reach that objective.

But we are underestimating the fact that AI soon will be able to do exactly the same thing. And so I think it is, it's a moment in time where it is very hard to feel like you're on an exponential, especially when you're further away from the technology. And, but we have to start thinking about the questions that you're asking, which is what does it really mean for us to live our day-to-day lives when AI is going to be doing the work that we might be doing today?

Linda Lacina, Meet The Leader: In your space, in the near term, right, so we won't even say five years – that just seems like years in the future, that's just Jetson-style stuff, right? Why don't we just say maybe in the next maybe 1 to 3 years, in your mind, what would be the skills that people should maybe start mastering? Like if you were if you were going to think about yourself and your team, what would you like to be having them shore up?

Eiso Kant, poolside: If you’re in knowledge work today, you need to assume that within three years there's an extremely high probability and likelihood that everything you are doing today in your role that is happening, let's start that's happening behind a screen because I think it's the easiest for us to think about, like the work we're doing in Excel and email and code editors, even in Zoom calls, that time is going to free up.

And so that often leaves the human element. The, the, at the end of the day, I like to quote, and I unfortunately can’t remember who said it, that, you know, AI is more capable of playing chess, but we still want to watch Magnus Carlsen play. And I think there is, frankly, parts of businesses and parts of what we do where that holds true.

I also think we just don't have the answers right now. And I think that's fair. It's very fair because when we are able to drive down the cost of intelligence, to the cost of energy and chips, we're going to have to rethink what it means to become economically valuable, to have that purpose and to be able to, what does capitalism to some extent even look like?

And I don't have all the answers today. And I also think that some of these answers shouldn't necessarily come from only business leaders. They should come from everyone, including in government.

Linda Lacina, Meet The Leader: You personally, what are you going to, what do you have your eye on as far as you want to either get better at or learn more in your mind?

Eiso Kant, poolside: So the reason I do what I do, a big part of it is that I love working with massively ambitious, kindhearted, low ego, extremely high intellectual horsepower people who deeply care about their work and their mission.

And so for me, I am deeply aware that we are, that the enjoyment of what we get to do on a day-to-day basis going after really hard problems is a human thing, and it's something we do together. So for me, the part that I always think about is what could I be doing today that I wasn't doing a year ago to really push that forward? And frankly, a lot of that has to do by just listening to the people around me who work to work with me to understand, hey, where can you improve?

Linda Lacina, Meet The Leader: Underlying so much of our conversation -- it's not spoken -- but underlying so many of the conversations here about technology and what people need to be prepared for is a fear that people will be left behind. Whether it's the staffs, whether it's the people who are already being underserved or even leaders itself, right? Because if businesses become, they need to stay relevant, to be going concerns, right? You know, what do you think about that? How do you how do you deal with this very natural fear of wanting to stay current and stay relevant and stay up to date in this very, very fast-moving landscape?

Eiso Kant, poolside: I think fear is, is, is the right word. I think I think fear is there as a mechanism to make us alert. And I think it's important, especially in AI, not to be yelling bomb in a theatre every year about AGIs around the corner. It's why I'm very careful with making the statements about 18 to 36 months for a lot of knowledge work. And, and I also, you know put that with the fact that I have an extremely high probability of that happening but is not certainty yet today.

And so, frankly, I think having the fear is already, is, is creating knowledge that it's there. And then secondly, if we know that AI is going to become an increasingly larger part of our life, of our work life and of our personal life, is to just spend time with it. Because as you're spending time with it, you're going to see the progress and you're going to be able to start answering, you know, parts of these questions over time in bits as you see, that AI is becoming a tool. And it someday maybe some of the fears that we have today are very warranted. But you can be part of the conversation then on how you want the world to look like, both to your leaders in business and in government.

And so I wouldn't ignore it. I wouldn't put the blinders on. I would just go step by step, use the technology and get a sense for yourself how it's progressing.

Linda Lacina, Meet The Leader: So, you know, you’re a co-founder, what was the moment that you, that inspired you to start poolside?

Eiso Kant, poolside: So I started work on AI. for its ability to write code in 2016, far before starting poolside. So I took a stab at this before actually, and we were too early. It's actually how I met my co-founder. He was the CTO of Github, one of the largest developer platforms in the world, and he offered to acquire that company. And, and he offered to acquire that company and we turned it down. But we became a great friendship.

And but the moment, really for poolside was that I had spent years of my life working on this problem and we didn't succeed. And then a couple of years later, ChatGPT comes out and it was incredible. It was instant within minutes of using it, that the future we always knew was going to happen was about to accelerate. And, and my co-founder saw exactly the same thing. And we have known each other very well at this point. We used to have a podcast together, which is frankly just a great way for two grown men to spend time together. And, and we started having the conversation of if we have such conviction, and at the time we thought it was 10 years of what the next decade looks like, what are we going to do?

And we realized we had our own points of view that were different from our competition in terms of how we wanted to tackle this, how we wanted to bring it into the world, and how we also wanted to progress the capabilities. And so at some point it didn't become so much of a question, it felt more like a mission. And I still remember that feeling. And we just got to work. And frankly, we're, we're still at it. And we've got a lot of work ahead of us.

Linda Lacina, Meet The Leader: In technology and innovation it's very, very natural and common to hit a wall. You think it's going to go one way, it doesn't go that way. You get through it. But there is a moment where you’re like I don't I don't know how we're going to get you to next. What was a moment like that for you, that was shaping for you?

Eiso Kant, poolside: So I want to separate the question because hitting a wall is often used in AI as, right now, the the easy thing to say, technology's hit the wall, I won't progress. And so, and I don't think that's the world that we are living in right now. But if I take your question from which I think your intent is of in your own journey, right, in your own journey of poolside, like, I never talk of hitting a wall, I talk all the time to the team about getting slapped in the face.

I think the reality of it is, is when you're going after something ambitious, the journey itself is hard and you're going to get slapped in the face. Sometimes there's months where it's just only on the up, and then the day you don't realise it, you get slapped in the face. And this and this will continuously happen throughout your journey. But if you've been around long enough and have tried going after hard problems before, and kind of in your question as well, you realize that this is going to continuously happen and the only thing that matters is that you're left standing on the other end of it, that you can get back up.

And, and that's really, you know, I'm not a big basketball fan, but Kobe Bryant said something at some point and he said there is no such thing as failure, the only failure is if you don't go back up and go at it again. And frankly, that's a little bit the sense that we've taken from this. It's an incredibly important mission. We have a ton of work ahead of us. We will get slapped in the face, punched in the face, even, probably along the way. We're in a race, frankly, or often I refer to it as a, as a bar fight, because AI really feels like that at times. And we're just going to keep chopping away at it. And the good thing is you build up a little bit resiliency over time.

Linda Lacina, Meet The Leader: So is there something that you do now that just would not have considered, you would not have considered doing even all the way back in the year 2016, something that you do as a leader?

Eiso Kant, poolside: Meditation. The reality of it is, when you're in this race, you're constantly on. Your brain doesn't stop, you wake up at three, four in the morning because you're thinking about X, Y or Z. But the reality of it is we're not superhumans. And I think we're all, you know, deeply flawed in our own ways. And the reality of it is we need to make sure that we can keep our energy levels up for this race.

And so for me, one of the things that ended up being very helpful both to make sure I could get my sleep in order, have the moments in time where the brain's actually clear and not running the whole day is something as simple as 10 minutes of meditation. If you were to ask me about this a decade ago, I probably would have dismissed it as some out there thing. And today I see it as a very valuable tool, combined with good sleep and exercise and good nutrition.

Linda Lacina, Meet The Leader: What is your routine?

Eiso Kant, poolside: So a big part of my role has me flying a lot. And so it's not unusual for me to spend, you know, several weeks a month on the West Coast of the United States and then several weeks a month in Europe. So jetlag is a big part of my life. And so I try to be very, very religious about my sleep from an hour-shifting perspective of every night, the same bedtime doesn't really end up being possible. We’re here at Davos, you have something where you need to be doesn't really, you know, lend itself well to a 9pm bedtime. So but I do try to make sure that my calendar is always structured in the way that I least have the possibility to get eight hours of sleep. That's not always possible. The brain, you know, keeps you awake. You might spend a little bit too much time on your phone, but so that's first and foremost. I think about my sleep.

Second is “no” to things that have an impact on my body and how I feel. And I think this is very different for everyone. But at the same time, you know, there's things that are quite universal. For me, making sure that when my energy levels are low, I make sure I have enough regular exercise. And making sure that the things I eat don't glucose-spike me.

You know, it's it's it's very little details. And I realised over time that we can learn a lot from athletes because at the end of the day when we're talking about trying to have top mental performance, right, because that’s fundamentally our role, you need top physical performance. And, and it doesn’t mean we’re all good at it. You know, we all have the times where we're jet lagged and we grab the candy bar and I'm, at times and, you know, I go three months without sugar, but then you pull up that chocolate bar. But I think it's working on that and constantly improving it.

And that for me was definitely a learning over the years that if you're trying to go after something that is a race that is incredibly competitive, it requires every moment of your attention and frankly means that you, you don't get to spend lots of time in other areas of your life, and that you're choosing for this, is just to make sure you do that. And then occasionally, you know, connecting off and cuddling up with my better half and our golden retrievers is incredibly helpful.

Linda Lacina, Meet The Leader: And our very last question because we only have one minute left, what is a piece of advice that you’ve always been grateful for?

Eiso Kant, poolside: I actually, I sent a tweet out a couple of days ago. My phone is in my pocket and I'll paraphrase it. That’s ok, I think I can paraphrase it. And I said something along the lines of early on in my career, and I must have been, I'm not sure if I was even 20, and I was looking at America and this was probably must have been around the time of the financial crisis. And I said, America is on the down. Maybe it's the end of the era of America. And someone sat me down and they showed me all the times in history when people thought it was the end of the United States of America or that it was on the decline. And they said, look, you have to understand that America is a country of extremes. It has extreme negatives. It has extreme positive outcomes. And the reality is, no matter when when it is on the down, it's a country of comebacks. It's a country of incredible, deep optimism of, of wanting to build the future. And every time we think that it's that it's down and out, it got slapped out, it comes back and it comes back stronger.

And I'm European with probably some European sensitivities at times as well, too, to America, but I got that advice and I took it to heart. And if I now look back over those last 16 years, from the financial crisis to COVID to the changes in in in in politics and presidencies, at the end of the day, America is a continent that is up and to the right and I think will, will be for a very long time because of the nature of, of this immense optimism of people. And I think the rest of the world can use a little bit of that American spirit.

And so that was some of the best advice I got. And I try not to forget. And I think it's very relevant in today's times because I'm flying to the United States next week and I know I'm going to be welcomed by a world that wants to go build towards the future of AI with a lot of optimism and the ability and willingness to take risk far more than anywhere else in the world. And I wish we could find the same here in Europe where we're sitting. But I think we, we're still a mindset shift away from, from getting there.

Linda Lacina, Meet The Leader: Thank you very much.

Eiso Kant, poolside: You’re welcome.

Linda Lacina, Meet The Leader: And thanks also to our listeners and our viewers. For more video podcasts, go to the World Economic Forum's YouTube page. And for podcasts and transcripts, go to the World Economic Forum's website at wef.ch/podcasts.