AI Won’t Kill Jobs - Here Is What You Need To Know | Ben Zweig | Revelio Labs

Shamil Malachiyev: Hello, everyone. Our guest today is Ben Zweig, CEO and founder of Revelio Labs, a workforce intelligence company that allows you to understand the workforce dynamics of your company — and, I think, of the industry overall as well. Hi, Ben.

Ben Zweig: Hey, how's it going? Nice to be here.

Shamil Malachiyev: Thank you for coming. Could you tell us a little bit about what you do at Revelio Labs?

Ben Zweig: Yeah, sure. So we're a data company, specifically related to employment data. We collect, curate, enrich and aggregate all the information in the world related to employment. A lot of that exists just out there in the public domain. There are resumes and LinkedIn profiles and job postings and Glassdoor reviews and all these different sources of information. We collect all of it. And most importantly, we standardize it. The data itself is a mess. As you can imagine — picture someone's resume. They write whatever they want. It's all free text. And that requires analysis to be able to get it into a workflow where you can do any sort of analysis on it. Titles have to be classified to an occupation, to a seniority level. Companies have to be mapped to an entity and an industry, and things like that. So we do a lot of data enrichment, ultimately, to get a dataset that's really useful.

And once it's all standardized and nice and neat, clients can use it to understand the dynamics of any company. They can say: at this company I'm tracking, maybe there's a high rate of attrition among their salespeople, and maybe that's a sign that they're not going to hit their targets. If I'm a competitor, maybe I see an opportunity, or try to recruit them. If I'm an investor, maybe I'll short their stock. There are a lot of people who want to understand companies — understand the inner workings of companies — and the best way to do that is through their workforce. That's the engine that makes a company tick.

Shamil Malachiyev: Yeah — I was just watching a documentary on Bloomberg a couple of days ago. The amount of data that they have access to is incredible. And I think one of the extra layers they could go into when analyzing a company would be the workforce trajectories — to see how things are really going, not just from the public materials.

Ben Zweig: Yeah, it's funny. I think Bloomberg the company is one of the most impressive companies out there. They really brought such sophistication to the world of capital markets, and we respect the hell out of them. They've done a great job in changing a whole sector of the economy and making it more sophisticated. But the Bloomberg terminal, which is so ubiquitous and so powerful? They actually don't have as much data as we have already, because capital markets are simpler than labor markets. The way that you can segment data, the flexibility that's required for financial market data — you don't need as much flexibility. There's a lot more rigidity around the way that financial market data can be analyzed. I remember asking them recently — I think they have something like 200,000 data series, and they store them all as pre-computed time series fields. And that's really a small fraction of what we do.

There are infinite ways to slice and dice the workforce data of even a single company. Let's say you want to segment by occupation, and there's a thousand occupations, and you want it over time — let's say 150 time periods. That's already 150,000 data points. And maybe you also want it segmented by country, and there's 150 countries. That's already getting into the millions. Maybe you also want it segmented by skill, and there's 10,000 skills. This gets into the billions and trillions of observations very quickly. And that's just for a single company. So the datasets can be really tremendous and massive, and just engineering this is a big effort.

Shamil Malachiyev: How difficult is it to get access to such large amounts of data, and to know how to structure it correctly to make use of it?

Ben Zweig: They're really different questions. Getting access to the data — there are big data-sourcing challenges. There's sourcing in terms of actually scraping publicly accessible data. Sometimes it's very hard to scrape it. You need a lot of proxies and a lot of technology just to collect data at scale. And you have to build in systems that make sure you're not losing data along the way, that you're not introducing errors, that you have consistency, and all that. So there's a lot of technology that goes into sourcing the data. Then there are anti-bot protections. Sometimes companies make it very difficult to scrape their data. They make it hard, or make it expensive, and it's a cat-and-mouse game where you try to get it in as efficient a way as possible.

Shamil Malachiyev: Companies that want to sell the data, and not just provide it for free, maybe.

Ben Zweig: Sometimes. Or maybe they want to gatekeep it. In the case of LinkedIn, it's public domain data, but maybe they want to produce their own products around it. In my view, that's anti-competitive — but I don't work for the Department of Justice; that's someone else's thing. So sourcing data is an issue. There are issues around privacy and all that — being compliant with GDPR. Lots of issues around sourcing.

But then, in terms of processing the data and structuring it and storing it, there are lots of things to do. Sometimes the data is just messy and it needs to be classified. I mentioned classifying occupations, companies, seniority levels, skills, work activities. All of that requires a lot of statistical models and large language models. There are also various biases in the data. There's sampling bias, there are lags in reporting — all of these need a lot of statistical machinery to make sure we're ultimately presenting data that is unbiased. And there are always improvements to make on these models; that has to be owned and maintained. Then there's structuring this data in a way that is not too computationally intensive or too storage-intensive. We have a dashboard that has a lot of flexibility, where we want to filter by things like keywords or skills. A lot of dashboards are built on pre-aggregating data, and they can run on a SQL backend. That's not feasible for data of this magnitude. So we basically need to run pipelines on the fly, which are more compute-intensive. And in order to do that efficiently, we can't use something like SQL or Python. We have to use lower-level, fast programming languages. So we build our own pipelines in Go and Rust and things like that. And then there's the infrastructure it's stored on — do we want to run this on AWS, or do we want to run it on our own hardware? So we have a lot of infrastructure considerations as well. But ultimately, at the end of the day, it's about being able to query information flexibly and easily. All that complexity — I think that's our problem, not the user's problem.

Shamil Malachiyev: And before, let's say, your obsession with data — you started your journey in a university. Could you tell me a little bit about that? Why university? Why the motivation to teach?

Ben Zweig: Yeah — so, I'm a recovering academic, as they say. I did my PhD in economics, and I love economics. The first economics class I took, I felt like I was walking into a new country and I already understood the language. That was the vibe. I was like: oh, this is how I already think — I can't believe there's a discipline for this. And it's exciting. The questions are interesting, the methodology is fascinating. And I also like teaching. I still teach — I teach the future of work at New York University, and it's great. I enjoy it. I think it's a way to stay connected to the deep thoughts about a domain. And even now that I've been out of real academia for so many years, I still feel like my real heroes are academics. There's an expression — I forget who said it — that if you want to know who you are, think about who your heroes are. And I still look up to academics, and I think of myself as an academic economist in my soul. So I was really drawn to it.

Shamil Malachiyev: And when you say you look up to people in academia — what do you think is the main difference between people who are in academia and people who found early success without completing it, or who dropped out?

Ben Zweig: I think if you're doing academic research, you are at the cutting edge of a domain. You are at the frontier of science — I mean, if you're good, you're at the frontier of science. There are plenty of people who don't quite get there, but you're at least on the journey to advancing our understanding of the world. And that is, I think, just a noble pursuit. And there are lots of noble pursuits that are not that — there are ways to contribute to the world and contribute to your family through different channels. But I think there is something unique about being an academic, in that you're trying to push the boundaries of science. And that's something I happen to be drawn toward.

Where I think I deviate from that identity — and I'm going to say this in a kind of negative way; I don't mean it to be as negative, so however I say it, cut it down by 50 percent in your mind — is that I think we're sold this myth that the way to advance a science is through academic channels, through this very specific channel of publication and peer review. And that is, I think, just not true. There are other ways to advance a science. Going back to what you were saying about Mike Bloomberg: this is someone who advanced the science of financial markets. Now, he didn't do it in the same way as Gene Fama or Harry Markowitz or other academic researchers who advanced the field of finance. He did it through a commercial channel, by creating ubiquity of data and products around analyzing finance. And that has been very effective. We've had academics and practitioners in financial markets. And I think we have not had that in labor markets. There are plenty of labor economists in academia, but the world doesn't have a Mike Bloomberg of labor markets. Hopefully we can get there. But I think we need that. I think we need more practitioners advancing the field of labor markets through products, and through platforms, and through things that scale differently — and, I would argue, better — than research.

Shamil Malachiyev: And at the same time, I think there is a question of financing. Because if you are exploring the cutting edge of any domain while at the same time building up revenue, you can always invest so much more into the areas that you find more lucrative or interesting, without having to rely on external funding.

Ben Zweig: Yeah, totally. There's a real tension there, at least conceptually. If you're not constrained by the profit motive, you can pursue interests that have more positive externalities, and all that. But in this domain of labor markets, I actually think there's not a lot of tension. I think you can advance science and also do things that have commercial value. I don't think that's always the case, but I happen to think there are enough ways to make money that also involve good, high-quality research. Maybe we're in a unique time, maybe we're in a unique market, but I think there's plenty of commercial opportunity and plenty of good, foundationally interesting work to do. There's a big intersection between those two priorities.

Shamil Malachiyev: What has always motivated you, and has your motivation changed over time?

Ben Zweig: Yeah, it's a good question. I think it has changed over time. I'll give you a little anecdote. When I was younger — in high school, and the early days of college — I wanted to do things that I enjoyed, that would make money, and support a family and all that. And I still really want to do all those things. Those are important priorities, and I feel like I'm mostly doing them. But I think at some point I got the itch to build something more — to do something that would impact the world somehow.

I remember at one point, my dad — my dad was a statistician, and I didn't really know that until I kind of became a statistician myself. I was like: oh, that's what you do? That's funny. But I remember one time he said something along the lines of: working with math and numbers is a great profession, because so many people hate it. If you don't mind doing something that other people don't like, you can make money that way. And I remember hearing that and thinking: sure, it's an interesting point — if you do something other people don't like, there are compensating differentials, you get compensated for that. But I also thought it was a bit of a narrow way to view your occupational choice. What we do in our job is kind of what we do with our life. It takes 80,000 hours of our life, and that's just really important. Sometimes I get annoyed when people push back against asking "what do you do?" You meet someone, you say "what do you do?", and sometimes you'll hear people say: come on, get real. But I think it's a really important question. This is how we spend our days. This is how we spend our time, and we should want to talk about it and be proud of it.

Shamil Malachiyev: Because we assume that most people, when thinking about what to spend at least 33 percent of their life doing, would choose something that they like, or enjoy, or would want to talk about.

Ben Zweig: Yeah, exactly. Something that they enjoy, something that lights them up, that they're passionate about — but also, it's an opportunity to build something lasting. And most people, I think, do build things with their careers, even if they don't realize it. You could work as a staff engineer at some firm, but you're actually building things that other people will build on, and you're creating goods and services for the economy, and those are things that people use. We're all part of building something that other people use and enjoy. That's happening whether people realize it or not. But when you do start thinking about it that way, it's kind of empowering. You think: I'm going to be allocating a third of my life to building something. So what do I want to build? What do I want the world to have when I die? That's just a fun way to think about a career. I like thinking about it that way. It feels empowering.

Shamil Malachiyev: And has your experience at IBM affected your vision of the world? Because I think that was your first experience of what happens at a very large scale, at a large technology company — which at that time was mostly focusing on data science, building smart cities, integrating data.

Ben Zweig: Yeah, I worked on a handful of those projects. So — at IBM, they had 400,000 people at the time. And I was so surprised by how hard it was to get anything done. In economics, sometimes we see a production function — the relationship between inputs and outputs. You can think of it as quantity of labor or capital on the x-axis, and the amount of goods and services you produce on the y-axis. And it looks like an S shape, where there's this inflection point where you get increasing returns to scale: the bigger you get, the more efficient you get. And that always made a lot of sense to me. There are, of course, returns to scale — you get better distribution and specialization and all that. And then it starts to get concave, and it tapers off: diminishing returns to scale. That's an intuitive enough concept too. But then sometimes in these diagrams, you would see an actual drop — output going down as inputs went up. We'd see that model in textbooks, and in the actual data. And it just never made sense to me. How could you actually get less productive as you scale? The cartoon example is a factory where everyone's bumping into each other and they can't get anything done because of congestion, and I thought that was such a silly, weird way to view the world.

But after being at IBM, it actually started to click. Having so much scale actually creates a lot of friction. There are managers on managers on managers on managers, and at the very bottom of this pyramid you have a few people actually doing work. With every layer of management there's communication friction, there are principal-agent problems, there's incentive misalignment, there's coordination failure, and there's territorialism. It becomes very, very hard to make any progress in a place where any idea can be squashed by some executive who's in a bad mood. That environment is just difficult. Now, I understand there are benefits to that sort of scale as well — at some point you have to adopt some sort of proceduralism and bureaucracy, and that creates room for specialization. But I started to get a little more pessimistic about the state of the economy. Part of the reason I went to work in the private sector was: all right, I'm an economist — I should see how goods and services get made in the real world. I've never actually seen stuff get made, and it would be great to see how a big successful company like IBM makes stuff. And after being there, I was like: wow, they're really not super efficient. They're slow.

But even though it's a little discouraging on the one hand, it also made me kind of optimistic — because there's a lot more to build. I was once getting lunch with Eric Brynjolfsson, an academic researcher who does a lot of work on the future of work. He was talking about conversations he'd been having with executives at consulting firms, and how they were worried about automation. And he just put this idea out there: does anyone actually think that we have made all the stuff that we want to make? Of course not. And it made me think: yeah, there's so much to do. Even at IBM — a huge, massive company — there were so many projects that we'd have loved to do with more resources and just could not. So that thinking started shaping the way I think about economic output in general: supply creates its own demand. Our economic output is constrained by how productive we can be. We're not going to run out of stuff to do. We're not going to run out of things to make. And it made me hopeful about what you can do as a startup. Just work on hard problems that other people aren't working on. That's all you need to do. The rest is commentary.

Shamil Malachiyev: And why do you think large companies don't just create spin-off startups? Instead of having so much stuff to do and scaling their workforce — which will make them even more inefficient, if we're talking about the diminishing returns on the number of people you have — why don't they just create innovation labs with ten people each, and do it that way?

Ben Zweig: Good question. I don't have a great answer. One way to think about it is: if you're someone who is drawn toward working in an innovation lab, maybe there is an option to do that at a company, but there's also an option to do it outside of a company. You could quit and start another company. And there are a lot of upsides to doing this outside the walls of a company. You get a lot more of the upside, you get more of the autonomy...

Shamil Malachiyev: And you experience the joys of fundraising and rejection.

Ben Zweig: Yeah, there you go — that's a plus, if you ever want to be totally bullied. Maybe you want to see how much abuse you can take. But even so, there is an excitement to doing something on your own. Now, that's risky, and not everyone can tolerate that kind of risk, which I totally get. But I think the pros outweigh the cons for someone who's very entrepreneurial — which makes me think that the best incubation spaces, the best ways to incubate, are probably outside of a real company. They probably get the best people. There's also a tendency for companies to — I don't generally buy into the idea that companies are super short-termist, but there is a short-termism that can come into play. If you're working on something and it's not paying off fast enough, it could just get cut, and that's a risk people may not want to tolerate. And it's very difficult to spin off a company that got incubated inside a large entity which still owns a meaningful percentage of it — it's going to be hard to find an investor who wants that overhang on the cap table. So I think there are plenty of inefficiencies relative to other startups. Having said all that, I do think it's underutilized. I think more companies should be creating incubated labs, or at least some experimental groups that are protected from the bureaucracy and structure of the big organization.

Shamil Malachiyev: You probably get asked this question a lot. Most of us — humans without the data — are looking at all this news about AI taking over jobs. With Sora releasing, my friends in Hollywood are thinking: whoa, what is this going to do? We're going to have so much content, so many movies. With you being able to see the actual data of how AI is affecting the world, correlated to the data of the past two, two and a half years — what are the main shifts that you're seeing? And is it making you more optimistic or pessimistic about the future of labor markets?

Ben Zweig: Yeah, it's a really, really big question — there are some sub-questions in there. First, I'll tell you how I think about technology affecting work generally, specifically as it relates to displacement of human workers. One way to think about this is that there's some speed at which technology gets adopted by firms. Firms adopt technology, and they represent the demand side of labor markets — firms hire workers. And that takes time. There's an adoption lag. We are seeing adoption, but it's not so fast, because firms are procedural and bureaucratic and all these things. So there's an adjustment period where we can breathe. There's also the time it takes for people to select into occupations. On the supply side of the labor market, there's a lag between when people choose what they want to be and when they actually enter the labor force. If someone could just switch occupations on a dime, we'd be relatively protected, but it doesn't always work that way. So a big question is how responsive labor markets are to technological displacement. Can workers reorient themselves, or does it take time? If lawyers get displaced, are all the lawyers just going to become, I don't know, psychologists? Probably not. So one question is the responsiveness of firms — can they react quickly or slowly? Another is the supply side — how responsive are workers?

And then there's a third part, which I think is the most important one in my mind, and that's the question of how quickly jobs can reconfigure. Sometimes the lazy way to think about AI displacing workers is to think about workers as being in occupations, and AI displacing occupations wholesale. But that's not how it works. Occupations don't get automated wholesale. There are very few cases in history where that's been true. You can make the case that switchboard operators were displaced in a moment, but nothing else really works that way. More often you have jobs changing — jobs reconfiguring. Technology usually automates tasks and components of work. A job is a collection of tasks, a bundle of work activities, and that bundle changes. And the way that it changes is an open question. I think it's one of the most important unanswered questions in labor economics: what actually determines the work that people do in their jobs? There are so many people — I know some personally, I'm sure you do too — who start a job thinking it's one thing, and it turns out to be something else. Their actual day-to-day looks really different. I think that depends partially on the technology, but partially on the organization as the organizing unit of labor. Is it the type of organization where you're hired into a predetermined set of rigid tasks? Or are you hired into a role where your job is to do whatever is useful, and it reconfigures all the time?

At Revelio Labs, we're a small company — only about 70 people — but it's very adaptive. Sometimes we'll get a client request which is a new thing, and we'll have to ask: who's got the bandwidth? Who could do this? Who's got the skills? Who's got the interest? And I feel like that's my job — I'm constantly thinking about how to reconfigure people's work. Maybe someone quits, and we have to think about who can take that on. Maybe someone says: I'm not interested in this anymore. We have to adapt, and we constantly reconfigure people's jobs to match the needs of the organization, which are evolving every day. And that has nothing to do with technology — that's just the day-to-day fluidity of being in a company with changing demands. Technology is a secular shift. But if you're adaptive to begin with, that reconfiguration is not a big deal — you're used to it.

So I like to think about it through the lens of job reconfiguration, conceptually. The tough thing is that we don't have a lot of data on how jobs are reconfiguring — the task composition of jobs. That's something I'm working on, and I've got a book coming out on the topic. It's called Job Architecture. It's about how we track the evolution of work, and how we structure our data so that we can measure these changes, track them in real time, and adapt to them. I think that's the big question for me. If we had more visibility into how work is transforming, we'd be able to be more purposeful in how we adapt, and we'd be able to do workforce planning in better, smarter ways. So these Hollywood studios — they'll adopt Sora, and maybe be able to do some component of film or CGI in a more automated, efficient way. But there's a lot to do to make a movie or make a show, and not all of it is going to be automated immediately.

Shamil Malachiyev: And maybe, if we become productive enough using all of those tools that we can do all of the week's work productively within three days, then we'll have more time to watch more films — which Hollywood will be producing in larger quantities using those tools.

Ben Zweig: There you go. In 1930, John Maynard Keynes wrote this article called Economic Possibilities for Our Grandchildren. The point of that article was to predict what would be possible in 100 years — and he wrote it in 1930, so we're not that far from 2030. It's right around the corner. And one of his predictions was that we'd see a kind of two-day work week. That we'll live a life of leisure — that was the main overarching prediction. That we'll be in a time of such prosperity that people won't need to work so much, and we'll live a life of leisure and appreciation of the arts and things like that. Now that we're closer to 2030, I don't think we're going to see it quite that way. He was wrong in some ways — in most ways, he was wrong. But there are some indications that he might have been partially right. I think work now involves more leisure. Leisure and work are more blended than they've ever been. It's not like we do backbreaking labor and then go home and enjoy ourselves — we enjoy ourselves at work. Right now, we're recording a podcast. This is fun. This is the type of thing that would have been considered leisure not so long ago, but now we're doing it as part of work. So I think that's worth pointing out. And we do see pushes for a four-day work week. Those haven't really panned out, but it's got some legs as a movement. It's easy to scoff at the prediction because it clearly didn't come true, but...

Shamil Malachiyev: A hundred years. A hundred-year time frame. It's not easy to make those kinds of predictions.

Ben Zweig: Yeah — it's hard enough to predict tomorrow. So we've got to cut him some slack. But either way, I think there's some wisdom there.

Shamil Malachiyev: Can we talk a bit about your experience as a CEO? Because this is the point where I hear so many different opinions and experiences, and I tend to lean towards there being a frame of mind that we need to have as effective CEOs. There are those that treat their company as "we're a family here." There are those that say: we're a machine that needs to get the work done. There are CEOs that suffer from trying to be nice, and trying to be seen as nice. How did you navigate your path as CEO? What were the main lessons — the mistakes that you noticed yourself make and learn from throughout the years?

Ben Zweig: Yeah — there's no shortage of mistakes and hard lessons. You know the expression that every general is fighting the last war? I sometimes feel that way — a lot of the way I've thought about building Revelio Labs is a reaction to my experience at IBM, where everything was very hierarchical and there was a lot of management. At Revelio, we don't have product managers or project managers. Everyone's hands-on. We are a hands-on company. There are really no layers of management. And that's a choice — it's hard to optimize around that paradigm, but we're trying to optimize around it. And part of the analog there — also a reaction to being in a big company — is that standards are really high at Revelio. We want it to feel like a high-performing team, with really high standards for performance. And I think that creates an excitement, in a way. People like being part of an elite team. The best people want to work with the best people. If you can create an environment where standards are really high, that attracts the right people — the types of people that also don't need management. The people who take initiative, take pride in their work, and keep motivating each other and pushing each other forward.

Shamil Malachiyev: How do you filter for those kinds of people, in interviews or the selection process?

Ben Zweig: It's really hard. I don't know. Interviews, in general — there's so much error. There's so much you can't filter for. I think we really do the best filtering after someone's at the company. I wish we did better filtering during interviews, and we've gotten better at it. But parting ways with someone who's not quite there is sometimes the move you have to make.

Shamil Malachiyev: How difficult is it for you to make that move?

Ben Zweig: How difficult is it for me personally to let go of someone? Not as difficult as it used to be. I think I've become a little bit more of an asshole. But I think I've been relatively transparent about it. We've gotten to a point where, if it's not working out, it's no longer a surprise — I don't think people take it as a surprise. It happens, and I think it's just part of the company culture that people understand. And of course, we do it as respectfully as possible. It's less difficult now. In the past, we used to flub this all the time — it would be without proper warning, or done in a reactive way. We had to learn how to be friendly people while also keeping expectations high. That's been a work in progress. And during interviews — I think the fact that people know we will pull the trigger on ending things if it's not working out has made everyone else who interviews more discriminating. They keep standards high, because they don't want to have to end things with someone who's perfectly nice. So it's always difficult, but it has to be done.

Shamil Malachiyev: I believe for a good CEO it's constant work on yourself. It's admitting to yourself that you are sometimes a little bit of an asshole — but sometimes you have to be, to make sure you can stand up for what you really think is the right thing to do for the company. Because internally you might want to be likable — everybody likes me, I'm a nice guy. But in reality, it's not the best thing for the business to have a CEO who has a problem being assertive in certain ways, making tough decisions, right?

Ben Zweig: Yeah, sometimes. I think if we can identify the reason for being assertive — justify being assertive, or aggressive — then at least we get a pass. There's an example that just happened a few weeks ago. We were at a conference, and right before, there was someone who we had collaborated on an article with, and then he went to work for a competitor and basically published this research as his own work, even though it was kind of a replication of our work. And I was upset about it — as someone who's kind of pseudo-academic, I think academic expectations and standards deserve some kind of respect. I had to basically call him out and tell him: hey, we're really not happy about this. I think this is not cool, and I don't appreciate it. I had to get aggressive. But really — it's not that I give a shit about getting attribution; who cares? The practical manifestations don't really matter. But I felt like, as CEO of a company, I've got to protect my team. There are people that worked on this, and if I'm not going to stand up for them, then how can I look myself in the mirror and pretend to lead this team? You've got people who rely on you to represent them, and to make sure they make the most out of their careers and their reputations. So I felt like I had to get out there and say it.

Shamil Malachiyev: There's a saying that I really like: it's better to be a samurai in a garden than a gardener at war. It's about suppressing your inner aggression while knowing that you have that capability — that energy to defend your team, your company, your vision, what's right, when you need to. It doesn't mean you have to use it all the time. It's having that capacity, and knowing you have access to it when you need it.

Ben Zweig: Yeah, interesting. I think there's something to be said for that. Having a reputation as a bit of a hothead is not a good reputation — you don't want to be someone who's going to fly off the handle. But setting that boundary — setting a line and letting people know that we can snap if someone crosses that line — is probably a good thing for people to know exists. So yeah, I hear you.

Shamil Malachiyev: Because I always think about the very critical side of this — like Steve Jobs, when people were afraid to get into an elevator with him, because he would ask a question, and if he didn't like your answer, you'd be fired. But at the same time, this allowed him to move things at such a speed, without ever being okay with mediocrity. He would always demand excellence, and you get that kind of speed of things running around him. What do you think of that?

Ben Zweig: Yeah — I wouldn't go so far as to say that there are necessarily really high returns to being a kind of piece of shit at work. There are some CEOs who manage like tyrants — "you're not going home until this is done" — and that can be toxic. And I think sometimes that gets romanticized. Having high expectations, I think, is really important, and sometimes there's no nice, friendly way to say that. So for sure, there's a spectrum here. But I also don't think there's such a tension between being someone that others do want to work with — and do want to get into an elevator with — and being able to propel things forward. Sometimes I'll call people on my team randomly throughout the day, and sometimes the conversation will end with: yeah, we've got a lot to do, I've got a lot on my plate. And sometimes the conversation will end with both of us feeling energized. We'll talk about ideas, we'll diagnose what's going on, we'll brainstorm. And at the end, we'll both feel excited — this could be really awesome if we get there — and we'll both want to move faster and do things better. So, I don't know. It's hard to motivate people, generally.

Shamil Malachiyev: But it's something that you feel you have to get good at to be a good leader — to be a mentor and motivator.

Ben Zweig: I do think so. It would be very hard to run a business without being able to motivate people somewhat. Now, I don't think you necessarily have to be this inspirational visionary — you can still be a practical, normal person. But I think it would be very difficult to build something impressive if you weren't able to get people to perform well. And I think the most productive people want to perform at their best. They're also drawn toward people who can identify what makes their work really good, point that out, and get them excited about problems that they might be less excited about if you hadn't had that conversation. Sometimes just getting into the weeds with someone and talking about things they want to nerd out about is motivating. We have someone on the team who's really into distributed systems. And I didn't know anything about distributed systems, but he was all excited about it — so I read a book on distributed systems. It wasn't that hard; it was an audiobook, I finished it in a couple of weeks. And now I'm able to have a conversation with this guy and talk about how these systems work. Just being able to have those conversations allows him to feel like: oh, I'm doing something that's appreciated. And that is sometimes all it takes.

Shamil Malachiyev: I can imagine it would mean a lot to people to know that their leader would take the time to learn in depth about what they're doing, to be able to talk to them — to make sure they feel valued. I think that already says a lot about your approach to leading. And what are you personally most excited about, looking into the future?

Ben Zweig: So much. I'm excited for labor markets, honestly. Now, we're in a little bit of a dark time in the US, where markets are kind of dissolving before us. But current events aside, I think we're on this long path where labor markets are really getting the attention they deserve, and getting more efficient and scientific. And I think that's just so exciting, and I want to be part of it. That was an original motivation for starting the company — I felt like this is a field that is advancing and becoming more scientific, more rigorous, and I want to do whatever I can to help get it there. And I think we're doing what we can — building as much as we can, as fast as we can, as well as we can. I really do feel good about what we're putting out into the world. I think we're able to structure labor market data in ways that it hasn't been structured before. We're able to derive insights about the effects of AI, or labor market health, or how companies differ from their competitors. These are important questions, and they're in the weeds and require a lot of hard work, but I feel like we're entering a world where those types of questions get appreciated more.

Shamil Malachiyev: And how are you personally making use of AI in your day-to-day?

Ben Zweig: In personal life, or in work?

Shamil Malachiyev: Well, I think it's both, really.

Ben Zweig: In work, we're using it for summarization, for labeling, for parsing — so many parts of the data life cycle can benefit from generative AI. And we really started this company in the early days of large language models. We were built on Word2vec right when it came out — this was like the first large language model. And now we're able to build on transformers, and on what's coming out of the generative models. So we've been able to piggyback on those advances in a way that's exciting — it gets us to be more efficient and able to do cooler things. In personal life, I use it all the time. I use it for recipes, and business strategy, and generating pictures of unicorns for my kids. Silly things. Even just ranking movies that are age-appropriate for my four-year-old daughter. It's so useful for stuff like that. Any time something comes up, that's what I turn to — ChatGPT or DeepSeek or Claude, any of these things. It's great. It's just making life a little easier.

Shamil Malachiyev: And to finalize — what advice would you give to yourself before you even got to work at IBM, with everything that you know now, everything you've learned about the industry, the labor markets, starting the company?

Ben Zweig: Wow, that's a good question. One thing that I would stress is that it takes a long time. I think business is, in important ways, a relationship game. You've got to build relationships, build a reputation, get to be known in communities, surround yourself with people that you want to learn from. And that takes a long time. When we started selling data into financial services, we went to a conference one year and people were like: who are you? What do you do? Okay, I'll probably never see you again. Then we saw them the next year, and they were like: yeah, I remember you. And then the next year they were like: okay, so you're here to stay. Now we can be friends. That takes years. So I'd say being deliberate about how you construct your network is really important. And there are so many things I would like to have known — even just how to consume information effectively, being able to talk the talk, getting to know the problems. Consuming podcasts is one thing that I didn't used to do, and now it's a huge part of the way I stay in the know about things. There are so many things like that which would have made me more productive — given me a head start.

Shamil Malachiyev: Can you focus for a moment on what you said about constructing your network? Maybe you can give a bit more detail on how to go about it.

Ben Zweig: Yeah. I once heard a definition of community. You could define community as where you keep showing up. And I like that definition, because it's so simple, but it's also actually practical. If you want to be part of the community, just keep showing up where these people show up. Go to their seminars, or conferences. Find a community that you want to be part of, and show up where they show up. That's it. It's pretty simple, I think.

Shamil Malachiyev: Just do simple things. I think that's one of the best pieces of advice that I've heard. In order to build something lasting and successful, you have to be okay with doing boring, simple things for an extended period of time.

Ben Zweig: Yeah — hey, you've got to grind, you know?

Shamil Malachiyev: Well, I want to thank you so much for joining me today, and for sharing so much. I wish you all the best with Revelio Labs. I'll be waiting for the release of your book — hopefully it's going to come soon.

Ben Zweig: Thank you. Its date is January 13th.

Shamil Malachiyev: January — okay. And I think it will be incredibly interesting for anyone who's listening to follow Ben, and to stay in the loop on what the labor markets are going to look like quarter by quarter. Because I think this is one of the most important and exciting topics right now — to be able to prepare, to see things in advance, and to make the right moves that will keep you on top of the wave. Thanks so much, and stay in touch.

Ben Zweig: Sounds good. Thank you. Really appreciate it. You too.