LinkedIn's Former CPO on AI's Impact on Product Building and the Future of Work

(rhythmic music) [SARAH SOULE] The AI at GSB initiative exists to prepare you all and others who come to campus to spend time with us to think both strategically and responsibly about AI and to develop an informed perspective on how to lead in an AI-driven world. So I am so delighted today to introduce Tomer Cohen, former chief product officer of LinkedIn and currently an advisor to the company, also a board member of Gusto and an investor as well. He is going to be in conversation with our own Jenny Steiger today, as we think about some of the topics that are top of mind for all of us in this particular moment of AI.

A couple of quick words about Tomer, who is a GSB alum, but before coming to the GSB for his MBA, he studied at Ben-Gurion University. After Stanford, he founded first a startup called FellowUp and served as an entrepreneur in residence at Greylock Partners before going to LinkedIn in 2012. And it was 2020 when Tomer became the chief product officer at LinkedIn. Today, again, in conversation with Jenny, he's going to share insights on how roles and industries are changing due to AI, what he learned leading AI transformations on one of the world's most, um, used platforms, and what all of this means for all

of us in the GSB community, particularly those of you who are entering or re-entering the workforce during this pivotal time. So please join me in welcoming Tomer Cohen in conversation with Jenni Steiger. [JENNI STEIGER] Thanks so much for being here. Is it nice to be back on campus? [TOMER COHEN] Always fun to be here. [JENNI STEIGER] Has it changed since you left? [TOMER COHEN] I come here a couple of times a year, always for a class or two, but it's always amazing to be here, and it's always like this.

Actually, we had our reunion last year. That was very special. [JENNI STEIGER] Nice. So you came to Stanford in 2008. You heard Reid Hoffman speak and then went on to spend 14 years building LinkedIn. I think I can speak on behalf of most people in the room when I say being early at a company like LinkedIn is really the dream. Your bio now leads with builder, and you're starting your next chapter. So was your bet on LinkedIn conviction or luck, and how are you thinking about this next chapter for you?

[TOMER COHEN] So I did become a LinkedIn fan when I was at the GSB, but actually I opened my company after, so I didn't join LinkedIn. Actually, I had no intention of joining. I was just admiring the company, and it was actually my first quarter, 2008 was my first year at the GSB. There was, across the street, the Stanford Engineering School had a lecture about social networks, and there were, uh, two folks on stage.

There was a younger dude in his 20s. His name was Mark Zuckerberg. And Facebook back then was all the rage. It was insane. It was the world's largest photo album. That's how it was called. And everybody was completely enamored by the Facebook story. I was one of the few that thought that Reid's vision was incredible. Because Reid was talking somewhat of a boring narrative, but I thought it was very powerful.

He was talking about how if you build social platforms online, you can really get so many incredible things done, both professionally and as a company. And I was one of the few who actually left saying, "Wow, that LinkedIn thing is-- that vision is pretty remarkable." Um, and then, you know, forgot about it, opened my own startup, and then as things would end up, had a conversation with who was back then, uh, in my role, was the chief product officer. And he asked me, "Hey, how would you build-" I was an entrepreneur. I was building in mobile. He said, "How would you build, you know, LinkedIn as a mobile product?"

I shared arrogantly what I would do, and he said, "How about instead of talking about it, you come do it?" And that was my kind of entrance into LinkedIn. And then when I think about, like, today versus then, you know, then when I was betting on LinkedIn, I think it was betting on me. I was betting on a company vision. [JENNI STEIGER] Mm-hmm. [TOMER COHEN] And now you're betting on a vision for the world in many ways. Like, what's happening right now is gonna transform everything: how you work, how you interact, how you're going to find your loved ones in the future is going to be powered by this thing. So in many ways, what I'm betting

on is very much like vision for humanity, the vision for the economy. [JENNI STEIGER] And do you have a vision? [TOMER COHEN] I have components of that vision. I don't have it for the whole thing. But I think there's a few things which are really clear about the change. In many ways, I think the biggest thing for professionals, and I'll put it this in-- as an angle coming, coming from a work environment, is that change is happening much faster than we're able to respond to it. [JENNI STEIGER] Mm-hmm. [TOMER COHEN] Uh, I think the biggest insight I had in the last year and a half was that we're not like. the way we are wired right now,

the way the best practices are built, are not built for this moment because they're assuming that things will stabilize. You build best practices, they last for a decade or two, and then change happens again, and now changes happening faster than best practices are being able to be created. So then what you're building really is you're building a new type of work environment, but also behavioral aspect where you have to move faster to just keep up with change. And that's not something that we human beings love to do.

We are adverse to change. We don't like change. Even the one of us who pretend to like change, like myself, like deep inside I know I have an adversity for change, so I'm working on it. But that's the thing that I think will be a big part of the vision is how do you build people's capabilities to actually handle it? [JENNI STEIGER] Yeah. I saw a stat from LinkedIn that said by 2030, roughly 70% of the skills used in jobs today will have changed. So you've said that the most at-risk group, and I think this is a little bit controversial, is actually not people coming right out of college, and it's not the more senior leaders.

It's actually potentially the people in this room who have a bit of experience but may, you know, just have enough to coast on. So what does this group get wrong about how AI is changing their careers? [TOMER COHEN] So when I- So I first saw- I've been talking about OpenAI actually here at the GSB in a class I was teaching since 2017. Um, I first saw GPT-4 as part of a small leadership group at Microsoft in the fall of 2022. This was roughly nine months before it came to market.

Um, and I remember back then the consensus was like, "Oh my God, early in career roles are gone." Because AI is really good at like the kind of the intern type roles, the analytical stuff, the kind of grinding stuff at the beginning. I've changed my opinion pretty dramatically on it, and I doubled down on it on LinkedIn, but we actually build a whole program for early in career talent. Because then I saw early in career talent, and they came very AI savvy. They were resourceful with AI, they had fluency with AI, they had agency with AI. Then I looked at my own talent, my own talent was very resistant to AI.

Not resistant in how they talk because they all talk a big game, but then in their day-to-day, they were not interacting with it. And if they did, they were like, "Oh, I'm using ChatGPT." Like, that's not a real usage of AI. That's just a common phenomenon right now. So then when you look at early in career talent, they were all about it. And there's a company called Windsurf. It was kind of partly bought by OpenAI, and then it was officially Google. But they're doing coding developments, and they couldn't-- didn't have enough money to hire very senior engineers, so they hired out of early career talent

with a few senior engineers at the top. And that talent has turned out to be amazing for them because they're highly malleable. They don't have best practices yet, which is a good thing. So they're learning as they go, and then they're really adapted to the ways of working which hasn't been invented yet. So for example, uh, in product development, you have your usual, the usual trilogy of like PM engine design. No such thing exists there. [JENNI STEIGER] Mm. [TOMER COHEN] They're all flexing across. And that for me is that new way of interacting and building.

So in many ways, I changed my opinion on early career talent, materially, given they're more AI native. I think the concern I have with this group, not specifically all of you, you're all talented, you all do really well. Uh, but the kind of mid-career talent is- you've built best practices. You're already at this notion of, kind of you have this adverse reaction to change, and your roles are gonna be disrupted massively. So then if you don't adopt that beginner mindset, or the role I left, a year and a half ago when I joined the GSB, and the role I'm potentially wanna enter is very different than what you know. And if it's not, you're entering an old world.

You have to think about the new world you want to enter. And how do you become a change engine, not somebody who is going to enter something pretty archaic? You don't want to go to a company that your old role still exists. You want to go into a company where your new role is going to be completely redefined. You want to feel uncomfortable. You don't want to walk into a situation where you know what you need to do. And for, at this level, it's very, very hard to do. So for example, at LinkedIn, when I did the change, I did

it at the top and at the bottom, and I wasn't sure yet how to do it at the middle. [JENNI STEIGER] Hmm. [TOMER COHEN] Uh, because they're kind of not needed, they don't have the expertise yet or the judgment yet, but they're kind of past the kind of early in career, and that's a risky place to be. [JENNI STEIGER] Tell us a little bit more about that change. So you changed the role of being a product manager to a product builder. What does that actually look like in practice? And I would have concerns that there would be knock-on effects, like a lack of connectivity or, um, perverse incentives, and so what does that-- how does that play out in an organization?

[TOMER COHEN] Yeah, it's a great call. So basically, the change I've made at LinkedIn has been we have the same, you know, functions as any other tech company: product manager, engineer, design. But if you take a step back, the idea was to create a new type of archetype called a full stack builder, and it can work across the stack. Now, you ask yourself, how does it all start? And it actually starts there because the role of a builder is actually quite simple. If you all start a company, you basically solve it on your own, you take an idea, and you bring it to life. But then what happens usually at companies, and there's no nefarious reasons for it, it's just, you know, very much inertia, is that every role gets stretched to a lot of mini

roles and mini functions. So the role of a builder becomes, you know, there's a problem, you research the problem, you design it, you spec it, you code it, you test it, you take it to market. But then the speccing side at LinkedIn, for example, becomes trust review, product review, design review, uh, security review, privacy review, research, and becoming-- looking at twenty sources to understand something. So that simple flow becomes extremely complex underneath. And then you hire people for every type of that sub-motions. So then you don't only just have process complexity, you have organizational massive complexity.

I actually have a slide. I mapped LinkedIn's entire, all the functions, all the org charts, and it's massive. [JENNI STEIGER] Mm-hmm. [TOMER COHEN] So then, like, it's not surprising, and it happens to every large company. It's not surprising that, like, building something pretty simple takes multiple teams, multiple weeks, multiple code bases, and it's really, really hard. But right now, you have the ability to actually collapse the stack back up. You can actually go back to craftsmanship. You can go back to meritocracy.

You can start building it from scratch. And that was the idea. Now it's not easy to do. So what I started, I started from the top, my own leadership team. So all the product executives that were all functionally oriented, head of design, head of PM, head of BD, and I made them all product area leaders. I started from the bottom, the early career talent, because again, they don't know anything. I'm training them to be what I want. So I'm training them to be the product builders that I want them to become. They're like this kind of Renaissance builder.

They can code, they can design, they can BM. And then the idea is to gradually make your way to the middle. A-and that's been the process internally, and it's still ongoing. I think it's going really well at the top. It's starting really well at the bottom. And then in the middle, we have this kind of pockets of teams that are showing this ability and so on. And then we are pretty much writing the playbook for it, and that's going to continuously change as we go. [JENNI STEIGER] And how is the ratio changing of technical talent to non-technical talent in the organization now that you're collapsing the stack?

[TOMER COHEN] So, my goal is to avoid even that question, because then like, ultimately, what happens when you have that model successful, the mental model I have is like Navy SEALs. So I'm moving from massive amount of soldiers to Navy SEALs who can just do a lot, and they're trained on a mission. So instead of everybody knowing one piece of thing, they're trained on a mission. They become really good, and I can assemble teams that work really well like that. They can actually tackle emerging priorities in the organization, really, really well as we go through that. So then, as an example, in teams that actually operate

this way, there's no ratio. We just ask, "How many folks do we need for this?" But there's no longer, "Oh, I need an Android designer, an Android engineer, I need a backend." That exists in the old world, not in the new world. Now, if somebody wants to push me on that, I'm happy to, 'cause if I were to push myself, there's still other roles. So for example, there's still system builders. Right now, we still have people who are building the system for full stack builders to be successful. Somebody needs to still build the infrastructure. So in a way, we have full stack builders. And then if we have system builders,

then we still have specialists. But then specialists are fewer and fewer. So the analogy I give for specialists is like playing at the symphony. You need to be amazing. Like, where before, specialists were, you know, a pretty big group. Because I don't need you to be okay. I need you to be amazing as a specialist, but I need you to be really good at going across the stack. [JENNI STEIGER] Outside of the product organization, how did you see AI permeate the rest of the organization?

Were you, like, the leader taking charge, or did you have peers who were also implementing AI solutions throughout their organizations? [TOMER COHEN] Yeah, it wasn't uniform. So I can tell you, because I've been early to AI. I was sharing with you the story. I was teaching here in 2017, a class with Rob Siegel and Amir. It was about product. I think now it's different teachers who are teaching it. And my thing was, "Hey, I'm going to teach how to do AI in product management."

And I was saying things that were highly controversial in class and was not well received sometimes by Rob. But to give credit to Rob, Rob liked the dialogue, so he kept it going for, for many years. And then, you know, every year I was showing, hey, AI is taking more and more of the stack, and more of the stack. And actually, you don't even know, but you're interacting with AI all the time. What you see in Netflix designed by the AI, what you see on your Facebook feed, your LinkedIn feed designed by AI. TikTok is all AI.

Like literally, you're always interacting with AI, you just don't know it. Now, at least it has a face in a way, or a name. It's interacting with you. It's called Claude, but it's always there. It's always been there for the last decade and a half. So for me, I've been on that train for-I took the pill a long time ago. And then I think in the organization it changed. So I was mentioning in the fall of 2022, I first, with a small group of leaders at Microsoft, we had Sam and Greg from OpenAI.

They demonstrated ChatGPT. All of us, actually, it was a fun moment because when they showed it, And this is- I'm sitting in a room, and I'm easily the least accomplished person in the room. They're all amazing seminal paper writers about AI, and they were demonstrating ChatGPT on stage. And the room was kind of looking at it but not very impressed. And again, this is incredible, but they all thought it was somehow like canned. Like when Scott sees a pitch sometimes, somebody's pitching a company to him, sometimes there's nothing behind the pitch.

It's just a pitch. So he needs to understand if there's a real product behind it or not. It was the same moment. It was like, "Is this real, or this is just a canned, you know, uh, question and response?" And then they basically said, "How about you try it?" And then the audience, a few of us started to ask it questions, and then the room became very quiet. Then you really realized something was different. It was, you know. English is my second language, and I never really understood the word eerie until that moment. It was eerie.

It was weird. It was odd. So then when I went back to LinkedIn, part of it was actually seeing how do we rebuild the roadmap from scratch with AI. And I remember asking, you know, walking my team through the change. We all walked through ChatGPT-4 and then saying, "Okay, what's it like? With this technology available to us, how can we reimagine what we're trying to accomplish for members and customers?" And I remember hoping for seeing dramatically different roadmaps. People came back with literally 99% of the same roadmaps because, again, people are adverse to change. And I flipped in a professional study, and said, "This is-- there's no way we're building the same thing when we have this tremendous amount of

composability behind us when we know we can literally accelerate." "So this is a no-go. Let's meet in a week, and I wanna see completely transformed roadmaps." And that led to that change. So it has to be pushed across. And something that my engineering partner and I did last year around the same time, we have every April, we have our R&D kickoff for the leadership team. It's roughly 300 people across LinkedIn. So we go on a summit. Usually my engineering partner and I walk through the priorities for the year.

We show the plan for the year. And this time we did something very different. We walked through a hackathon for the leadership team, and we have them all do 15 hours of mandatory vibe coding. Now this is some of the most technical people in the world that I am forcing to do mandatory vibe coding, and they didn't like me, to say the least. And I could see who was doing it or not doing it. There was an assignment. But by the end of it, after we did the hackathon, I had people who were basically building tremendous technologies for years saying, \"Until I did it, I didn't realize it.

I didn't realize how powerful this is.\" So they could-- were able to talk the talk, even work on the job, but weren't internalizing it until they actually felt it in their hands in a way. And that, that's kind of the pace of change you have to walk through. That's why, in many ways, when you see AI-native companies, they have an edge because they're coming without too many best practices or without some thinking around like what used to be done in the past, and they're just immersing themselves in the future. [JENNI STEIGER] Hopefully that's some good motivation for everyone to sign up for the AI at GSB Hackathon in May. [TOMER COHEN] It's mandatory.

[JENNI STEIGER] It's mandatory. I'm sure if you came back and taught your product management in the world of AI class now, it'd be oversubscribed. I want to talk about how companies are incentivizing AI usage, because at first, a lot of companies were saying that they would. that you would be valued based on how much of AI you were using, basically, like the token use, which of course got very expensive, and the more you use AI doesn't necessarily mean the more value you bring to the company. And so there's some strange incentives there. And it becomes almost some like unbounded variable expense for the company.

It's really hard to predict how much people are gonna use, and that you can't measure the value from that. So tell us a little bit about how you think that we should be measuring how people are using AI at work. [TOMER COHEN] So this is a key question I think every business leader is either grappling with or should be grappling with. So, so early on, actually at the beginning, very few were kind of token maximizers in a way. There were. Again, Most of them were still adverse to change. So they were still using, yeah, talking about it and getting excited, and they'd be sharing an article on LinkedIn, but weren't really using it. And very few were basically taking off

and using this technology. So you wanted that behavioral change. So you were, in a way, kind of nudging, encouraging, sometimes kind of mandatory, pushing people into using tools so they can really appreciate the technology and walk through it. Now we're past, I think for some companies. Again, I'm sure there's a lot of companies who are still going through that phase, but let's say for the top companies, I think that's very clear right now. So now you're in this phase around, like, Tokens are not value. Tokens could be tremendous value. Tokens could be a waste. Mm-hmm. And tokens are expensive. So how do you walk through that change? Because you're no longer in that, "I need you to use this."

I need you to learn. People are over the hump. Now you're like, "Okay," like, I need you to actually produce outcomes from it. So that, and that's, that's the real kind of change management right now to be done. And I think there's a few practices that are being built. As I mentioned, best practices are being built all the time, but what's happening right now is for kind of tech companies, they're building kind of some layer of excellence in the middle or governance in the middle, like an AI traffic control in a way. Something you can basically start saying, "Okay," you as the end user of an AI and a company, you don't decide what model you're using.

Like, nobody needs a Ferrari to drive to the grocery store. It's a waste of, a waste of a car. But if you want to race and win, you might need that car to go along with you. So that decision sometimes for companies is being done for you. You don't get to decide which models you're using. Some, for example, we had a lot of success with small language models. You don't need large language models for everything, so it depends on what you're trying to build. It becomes very tuned.

You're starting to bifurcate into very specific nuanced aspects of how it's getting done. But I would say the biggest thing has been kind of an ROI evaluator, moving from consumption to outcomes, from consumption to impact. So in a way, the way to think about it is almost like a funnel. At the top you have input metrics, and that's the stuff you control for at the beginning, like how many folks are using, how many folks are adopting, what's the tokens per employee? All of those are being measured at companies. Then you move to, okay, how many PRs are being done? How many experiments are we running right now?

Like in some areas of LinkedIn or high-growth areas in the company, the number of experiments usually is correlated highly with-- is causal to the level of revenue you'll see. So we know that's a strong push. But then at the bottom funnel, you start to look at like stuff like, okay, what's the revenue growth? What's the growth in engagement if you have an engagement-type company? What's the growth in member growth coming out of this one? So you start to measure that across, but that's like a new capability to build into it. If you're still stuck in the usage, then it's mostly a show.

It's not yet, it's more of theater. It's not yet value. So you need to start moving yourself down the bottom funnel to see if it's actually being measured and done really, really well. Now, it's not easy for, like, everybody. Sometimes I'm giving you, like, an abstract framework, but really, like, if you're the GSB, how do you measure value coming out of the usage of AI at the GSB? Not easy. So then, like you kind of walk through that objective function as you go through it. [JENNI STEIGER] Yeah. And for most large companies, the largest cost center is people, but now that cost center is being made up of people and compute.

So what does that mean for how organizations are structurally evolving? [TOMER COHEN] Yeah, in the future. So this is a really important distinction. So if you think like every CFO right now, and there's a great class about, there's a great CFO class here. [JENNI STEIGER] It's hard to get into. [TOMER COHEN] So actually, but there's a, I don't know, there's a few actually, and I think there's one coming up, uh, that Jeff is teaching, and I think the CFO of Meta is coming in. The best question to ask her is that question. How do you move from people, people plus compute? Because compute is ex-- and compute is, like, highly correlative for Meta's revenue because ad is highly dependent on AI, and that's not the case.

It's been the case for many, many years. So it's not something they're grappling for the last three years. It's, like, been a decade plus, but they know. And actually, they almost have it to a formula. So Google and Meta are probably the best ones in the market understanding how to translate AI usage into revenue usage because they deal with ads, and ads are highly dependent on AI. SaaS is harder. So, um, so people cost moved to be people and compute cost. And that's why you basically still have CFOs trying to grapple with, like, "Okay, it's not-- I need to invest in AI, it has to come from somewhere."

So there's the whole cost of people cost coming through the door. And I think that's the notion around that they're trying to work through in terms of the usage there. Again, if you have a great understanding of your output metrics, you can be a lot more informed in how you make those decisions. If you don't, then you're stuck because you know you have to invest in compute, but at the same time, you have not seen the market. And in many ways, the stock market reflects that today. A lot of companies are selling, companies are buying, but they're still not showing the impact on their own growth or even the promised impact on GDP.

Now, as an optimist, that will happen, but it's not that you can clearly see the line and you're saying, "Okay, that's going to happen in the next three years." So the bottom funnel impact will happen, but still pending some movements there to actually make it happen. So that's a big change. So you have to work through that change really rapidly. The other thing, it's not uniform. So right now, I would say like coding two years ago was a big promise with AI, was clearly going to happen. A year ago, the best companies, the best talent was using it. Now it's go, go, go, right? Like, it's hard to find somebody who's saying, "I don't know about Claude.

I'm not sure. I don't know if Codex is very good." Like, it's like a go, go, go with those tools. And so you can clearly see the value in the output, and that's in engineering. Design, not so much. Most designers, they'll talk a big game about Figma Make, but then you press them against the wall, they're still using the same best practices they had before. Marketers, similar. So it's not uniform. So if you lead the company, you have to think about where do you press. Right now, you press hard on engineering.

You should see tremendous throughput for that. You start to press on areas where you start to see the fruits coming out. But some areas you don't know how to press yet. So like it's not uniform in the way you approach AI usage and where do you actually apply compute to. [JENNI STEIGER] And you mentioned SaaS, so I'd be remiss if I didn't ask you about the SaaSpocalypse. What do you think about the idea that enterprise SaaS is dead? [TOMER COHEN] I like the provocation of it, because I think it pushes people to think differently about what they build. But the question is, what is SaaS used for?

If you are delivering outcomes for people, then nothing is dead. People need outcomes. If your whole existence is life is workflows, I'd be worried about that. Workflows are gonna transform materially. So it's not SaaS per se. SaaS is just a mechanism. If you have a company, if you're able to take the user all the way to productivity gains and they can measure it, I think you're in a really good spot, but if your whole existence-- you don't, if you don't have any unique data, if your entire existence is building pipes, unless those pipes are extremely unique, I would be worried about that thing becoming a moat.

Now, there are also a lot of those SaaS companies are in a phenomenal place to convert their own customer base to using the new solution. So it's not that they have the ability to compete. It's just that now their moat is at less stronger. But I would say, if you have unique data, if you have unique access to information, and if you think about right now, even if you build-- if many of you, especially in the second year, are thinking about, "Do I build a company or not?" I know you're building a company. So then there's the question about, "Where is my moat?"

"Right?" "Like, if the models are getting so, so much," you know, they're so strong and they're so capable every year, where do I build my moat? And that's not an easy question. How many of you know of Harvey? Wow. Yeah. So Harvey started by being a model wrapper. There's no moat there. But over time, they realized, wow, in legal, the governance is so, so hard, right? The ethical boundaries. Two partners cannot know about each other's case because there might be a conflict. So they start building like dedicated, unique AI solutions for those. That becomes their moat. So it's not the capability of like, wow, it's really

intelligent when it talks back. That's not your moat. Your moat is like all the attributes that make it unique for that category. And then again, you still have like an entropy goal of that. Yeah, of course, they can go after that, but they have a bigger pie to go after. But that's, that, that's your moat to go after. So I think that you start asking yourself the question around like what's, what's expected to be kind of, you know, grow, extend, formulate from that and what's not? And then you can start seeing those examples coming. Even the ROI calculator right now is a really important area.

Like there's many companies investing. The AI traffic control. All of those aspects of the stack are really important. So, like, in my mind, there's gonna be a lot more companies being built around those solutions. [JENNI STEIGER] Yeah. And you, when we spoke, you drew a line between AI fluency and AI agency. So a lot of people know how to talk about AI, but not many can actually tell you how they are re-engineering how they work. And so what's a tell for you that somebody's talking about it but doesn't actually know? [TOMER COHEN] When I asked them to show me something they built. So actually, I'm not going to ask for shows of hands, but how many of you use an agent today?

Everybody's using Claude, ChatGPT. They would be very surprised if somebody here doesn't use it today. But again, no need for show of hands. But how many of you have actually built something that automates something in your day proactively, that takes away something that you used to do every day proactively. Okay, so let's say 15%, which is quite high. Actually, even 10%, that's quite high. In most areas, nobody has done that. That for me is agency. Using ChatGPT is not agency. My daughter is seven years old. She uses ChatGPT, but an AI native, she's gonna be amazing. But that's not- But the notion around, like, how do you actually use? Because the components are there, the building blocks are there.

How do you start using them to make your day better, to get your job done better? That's agency. How do you lean in? So when we did the-we call them APB, Associate Product Builder program instead of the APM program, their interview was, they had to submit a product they built, no longer a resume. And then in the last round, we had engineering, product, and design coming to interview them one setting, and they had to build something from scratch in the interview. Talk about pressure. But it was a long time, and we were very compassionate, but the idea was to see, can you actually do it? And then I watched all the interviews. And it was not hard.

It was hard. It was like an hour and a half interview to watch. But, um, but it was amazing for me how they use those tools. Like, I was learning from how they were using tools. This is junior talent coming to the company. I'm the chief product officer. I'm learning from how they're using tools, like, bring them into LinkedIn. The way they were moving around the tools, the way they were basically automating the workflow, we basically said, "You bring your laptop, you bring whatever tool you want.

We give you the problem. You solve it any way you want." And then the way they were using it, for me, was mind-blowing. That for me was like, oh my God, that's the talent we wanted to bring into the company. So I think agency for me is you actually walk the walk. You start to use those tools proactively. You start to think about pains in your day that you can start automating. Even if your professor is telling you have an issue with something, you go and solve it for them in a way which is sustainable, you know, it continues. It's not like a one-off and so you gave them. That's agency.

[JENNI STEIGER] Mm-hmm. And so one way to think about a job is a bundle of skills, and a skill is a bundle of tasks, and tasks are what's really easy to automate with AI. And you've said that the amazing thing about the GSB is that we focus on soft skills like entrepreneurship, and we take Touchy Feely, and that's our strength here. And people in the room are making decisions about where to bet their futures and spending a lot of time thinking about where to build skills and where to dive in, and it's very uncertain right now. So walk us through how you would think about where to spend your time if you're an MBA 2 right now, and you're going to recruit in the next couple of months,

um, and where will AI be dominant? Where, what should we forget about or just push aside for now? [TOMER COHEN] So I would say if I zoom out, what to learn and what to teach is one of the hardest questions right now. Like, you should have a lot of empathy for your faculty. It's a hard problem. Like if you-- and I have for you as well, but like what to. what to learn? Like I ask, I have a podcast, and I bring amazing builders from the co-founder of Pixar to CEO of Spotify and ask everybody, like, in the end, like, "What would you recommend to people who want to go and enter your space?"

And I get very different answers across the board. Um, and I think it's hard. It's hard to think about what is it that you want, to learn or teach for people to become very successful, because knowledge is no longer unique, right? It's like there's no scarcity in knowledge. You can access knowledge. When I was here and I was learning a lot of amazing information, and part of it was frameworks and spreadsheets, and it was super useful for me. I don't think you need to memorize spreadsheets and frameworks any longer. I think hard skills will be taught very differently. If you're an MBA 1, before MBA 2. If you're an MBA 1, I would double down on soft skills. For me, it was by far.

Actually, I learned a lot of different hard skills and soft skills. Soft skills was. not my strong part. when I came to the GSB, and I've learned a lot. You know, actually, I got my worst grades in soft skills, but I learned so much. My best case came from hard skills. But I've learned so much through soft skills. And things like. things like empathy, like, how do you understand, how do you profoundly understand an unmet need? Things like creativity, how do you think beyond the obvious? Things like communication, like how do you rally organization is really, really hard.

I've learned this the hard way multiple times. I almost lost my job multiple times because of how to rally organization in a specific way. It took me a long time to learn that. Judgment, probably the most important skill for a builder. It's how do you make decisions in uncertain, ambiguous ways, like ambiguous situations. That's a really important skill. And I would say even on the hard skill side, if you're learning finance and strategy and accounting, is how to use those skills so you can have better judgment.

Not so you can do accounting, but you have better judgment around accounting or finance or strategy and so on. If you're in an MBA too, you don't have much time left to learn a lot. But I would say like then, I would basically try to find the best place to learn from outside or start something you feel really strongly about. I would not join an old world company because of a title or comp. You know, I left the GSB. I had a massive debt I wanted to pay as soon as possible 'cause I was never in debt before I come into the GSB, and then I had to take a big debt to come in. And then I decided to do a startup, and I credited the GSB,

not for telling me to do a startup, but to basically ground me what matters to me a lot, and it wasn't ultimately, you know. It was following what I believed in, which was a lot of what the GSB was pushing me on. So when I think about the GSB, I think about a lot of the teachings here. It's shaping you to be a better builder. Entrepreneurship is-- it's almost like it's, it's in the air here. It's really hard to avoid.

You feel it. Even if you're not going to start a company, like, a part of you leaves with entrepreneurship built in. Either you're gonna be that in a company or inside a company or outside, but it's kind of in the air here. It's pretty amazing. Those are tremendous assets to bring to the economy, and I would double down on them for the future of the GSB, and I would double down on all those human-specific, unique assets. Vision. If you're an MBA, too, go learn from amazing people. Literally. Put aside the comp. Put aside the titles. Titles don't matter.

Go learn from amazing people and things will just work out. And that's how you make your bet if you're an MBA2. [JENNI STEIGER] You mentioned judgment, which I think is really interesting. We were all pre-AI analysts in this room, and so I remember having to learn Excel so I didn't have to manually go through 20,000 cells of data and clean them up myself, which I did do for a little while until I learned how to use Excel well. And so I think that process of being really deep in the grunt work and understanding why things are done a certain way and how they're done is how you build that judgment. So if a lot of that grunt work is automated and entry-level,

people don't have to do it anymore, how do you build the judgment to eventually lead an organization or lead people? [TOMER COHEN] Yeah, I think the judgment comes a lot from honestly making mistakes, seeing others interacting. Like, that's where you kind of learn. Like, it's called intuition, but intuition for me is just like looking at past experiences. You just can't connect the dots, but your subconscious can. Um, so in a way, this is where, like, for me, if you wanna get great judgment, like if I think about like, people ask me, like my mentors.

I had so many mentors in my career because I picked up a small thing from everybody. Um, you know, somebody in an interaction, like I could see. Like I was-- you could see an interaction at work where, like a meeting is about to burst and somebody's able with one sentence to bring it back. That's a really specific judgment in the room. You know, this is not going anywhere. It's gonna be run really poorly, and somebody's able to bring it back. You pick up on that.

You run projects. You run experiments. You're so bullish. This is gonna be amazing. You don't. It doesn't work out. I did my startup. I earned nothing for a year. I was on an F2 visa. I had. It was expiring in 14 months. Had to go to Vancouver for a month because my visa was expiring. It wasn't pleasant, But the school, uh, I learned during my company was just unbelievable, was priceless, and the company was not successful.

This is not a success story of grinding through. This is a grinding through without a success story in the company, but if you think about my skill set growing through there, it was amazing. And then mistakes I've made, a lot of it because not listening to people 'cause I was kind of charging. But there's so many amazing leaders around you. Your faculty has gone through tremendous. Like, some of my favorite times at the GSB was doing lunch with a professor and learning so much. Um, that's judgment. That's how you start to build judgment. And then gradually you do it yourself, you succeed, you fail,

but then you kind of keep going, and that's how you build it. It's hard. This is when I talk about, like, the APB program- That's where they lack, but that's completely understandable, right? They need to learn this as they go. So you can't expect them to know exactly if this is a good product or not. They know how to build fast, but do they know how to build it well? Not yet. They would with help. [JENNI STEIGER] Right now, there's a lot of noise in the world and venture, and there's a lot of money flying towards AI.

Everything is AI all of a sudden. Everything is.ai all of a sudden. Um, and so you mentioned entrepreneurship, and this is obviously, like, the center of entrepreneurship maybe in the country. Um, and so do you think it's a good time to build right now? And if you were to go build something, What are the exciting areas for entrepreneurs? [TOMER COHEN] I think it's an incredible time to build right now. In a way, less than one percent of the world's population can code, much less than one percent. So in a way, that area was kind of pushed away for people. And in fact, when I was at the GSB,

I was like, "Can I find a technical person from the engineering school so I can do a company with?" That was a lot of the talk. Like, and I think that- [JENNI STEIGER] They don't let us over there anymore. [TOMER COHEN] They don't. There's no. there's a line. There's a fence. Um, a lot of that is moved away. Like, if you wanna. Actually, if you wanna prove you have a real product right now, you don't need. Again, like, all you need is a mindset.

The full stack builder mindset. It's great. Partnering with somebody is amazing 'cause it's easier to do as a couple or a trio than to do it alone. But in terms of like, "I don't know how to do this." Where in the world was there ever technology that taught you how to use it? It's just unbelievable. That idea that, like, if you don't know something, you just ask it. It's just incredible. So, um, I think, like, it's an incredible time to build and. But you have to be very thoughtful about how you go about it. Like, this is where I would start studying the different models out there, models of companies out there, and seeing, you know, like, where do you start building moat?

Uh, you start, like, what was the best practice around how fast technology was headed? That's changing massively. Um, like, you know, when I was talking about the room when we saw GPT-4 for the first time, again, I was sitting with some of the top AI researchers in the world, and their jaw was dropping because they thought this would come in 20, 30, 40 years, and suddenly it was in front of them. So whatever you think changed, how fast you think it will be, it's gonna be faster. Um, so then you have to, like, apply for that coming in. But this is an incredible time to build.

The last thing you want is to be in a situation where you're not progressing fast. That's why whether you're starting a company or joining a company, if you're starting a company, you're gonna learn fast. That's guaranteed 'cause you have to. If you're joining a company, please pick a company where you're gonna be uncomfortable and at the edge of learning all the time. Because if you're joining a company where they're not pushing the envelope, they're not thinking about this, then again, you're going to be building for old problems with old techniques, and you're going to make yourself not as useful for the economy. And I think that's a shame given your capacity and your potential.

So put yourself in a situation where your learning curve is going to be amazing. [JENNI STEIGER] That's some good parting advice. Before you leave, I'd love to open up for some questions. We have some mic runners, so if you have a question, please raise your hand and our mic runners will come find you. Oh, I think there's one in the middle here. Mason?

Oh, thank you, Mason. [STUDENT] Thank you for being here and such a great insight for both MBA 1 and MBA 2. So personally, I have a thesis on outcome-based pricing that you mentioned. [TOMER COHEN] You're MBA 1? [STUDENT] Yes, MBA 1. I have a thesis around the outcome-based pricing on how that changed SaaS and people start to like, um, see the pattern around that. Like earlier, you kind of briefly touched among that, but when I talked to a lot of the startups in the area, a lot of the time they face issues on how to.

How can you actually measure and track the values, the outcome to the AI SaaS? And I'm not sure, do we have any point of view on that? [TOMER COHEN] Yeah, so I think, you know, right now, in fact, we see a lot of companies when they started off, On the SaaS side, they were not sure how to price it. So then they priced it, and they added, uh, kind of additional cost to a seat. But the seat-based model is being challenged as well if you're in SaaS, right? Because you don't need as many seats 'cause the whole idea is to make the seats very valuable so that. So then that's a very open question

out there that I can tell you everybody's dealing with. Um, on the ad side, um, I'm gonna cheat for a second. It's very simple. Because the ad side, you deploy more compute, you get more revenue. Because you deploy more compute, you get more relevance. You get higher CTRs, click-through rate. You get better CPCs, you get better ROI. Like, the ad side is a goldmine right now. That's why you're seeing Google benefiting both from the ad side of the world as well as the fact that they have the full stack for the AI, but also Meta. And then on the SaaS side, you'll start-- It will hurt-- it will have to start moving away from seats.

The problem right now is there isn't a clear way of how do you measure outcomes. And even, by the way, if there was, there was gonna be so much debate about it. So it's not an absolute truth. I can show a marketer, "Hey, you got 10% more clicks." I've done my job. You go convert them. I can't-- like, what is it to tell a company, your employees are 10% more productive? So it's a big question happening right now. So you have to connect-- a company internally has to connect it all the way from input to output. They'll have to do the work to see the output and how that works, you know, all the way through. So part is like the company offering the product, and part of it is the company

itself in their own toolkit, showing the value internally. It will move to pricing per outcome, but that's gonna be a journey. But it's gonna-- there's a, a great saying that a lot of tech revolutions come with a business model with them. So I think the tech revolution has came, and the business model is forming, but it's not yet fully mature. [VANESSA MOORE] Hi, Tomer. Over here. Vanessa Moore. Thank you so much for being here. My question is about as the APB program is focused on speed and prototyping and learning how to build really fast, do you worry- [TOMER COHEN] And well, and well.

[VANESSA MOORE] And well. Well, that's the core part of the question. I'm wondering how, as speed has become such a priority, how you teach entry-level product managers to maintain that rigorous product work. Yeah. And I've been hearing at other companies that part's being lost and the experimentation culture is being taken to an extreme. [TOMER COHEN] We're dealing with this right now. So the APB Burger, for what it's worth, it's a baby. It's three months in the making.

We took six months to find the talent, and then it really started in January, so it's three months in. But we're dealing with exactly the same thing. This talent is really quick. In fact, you kind of, uh, in many ways, you're, you're precious of them because you don't, you don't want them to be feeling, "Oh my God," big company, I can't move." You want to make sure they feel supported because they're supposed to be change agents in the company for the years to come. But at the same time, like, they're shipping fast, but are they shipping the right things?

So then you kind of want to make sure there's like a good. I wouldn't call it a buddy program, but like, a good mentorship where there's like the velocity of what's coming in is really good. And in fact, if I were to zoom out, right now, when coding is becoming so fast, what you'll start seeing is design will be a bottleneck. Then design will become very fast. Then good ideas will become a bottleneck. So what I'm seeing right now with the APB program, it's honestly just kind of like a micro sample of what's going to happen across technology. Good ideas will start becoming bottlenecks because you can push so much, but are you pushing the right things? And then tied back to, are you seeing the outcomes from it?

So we are in this continuously evolving, I think there's a desire right now to have like a, 'Let's get a best practice playbook. Let's put it in place. Let's start kind of, you know, grinding through it.' Like LinkedIn is a massive best practice company. We love best practices, and part of the change was no longer stable best practice. Establish, move on, establish, move on.

[JENNI STEIGER] Mm-hmm. [TOMER COHEN] And if somebody's waiting for a best practice to happen, that's gonna take a long time. I don't know if it's worth waiting for. [JENNI STEIGER] Mm-hmm. Thank you. [EITAN] Hi. Thanks for coming. I'm Eitan, MBA1. How would you structure the organization of a young company today? And where do you think the equilibrium will be for today's multi-thousand people companies?

What will they look like whenever there's some sort of stability? [TOMER COHEN] I would start very easily. If you think about, like, if you, if you build a company right now, you only need the one person, somebody who builds the product and-- But if you have two, bring a go-to-market person. It could also can be full-stack go-to-market person. They can do. But what I said about R&D, exactly same thing for sales and marketing. Sales and marketing is super complex. And if you really press on it, a lot of it was just reduced because of all the sublayers, but there's no reason for that. So, so you kind of start builder at one, have two people go to market, then gradually depends on

where do you see your biggest capacity, uh, limitations and where do you want to start growing into? Um, I do see the future being a lot more of the system builders, full stack builders, a few specialists, and then more of the notion of like quick forming teams, like pods coming on the fly. I can't tell you how many times at a company you have an emerging priority and everybody says, "Let's go." And then it's been a week, and nobody's going anywhere because you have to meet people around, and you have to, like, close this and notify people and, uh. And it's really, really hard.

Really, you can't underestimate how-- overestimate how hard it is for a big company to move and form. This simplifies it materially. Really. Now, it does also create some kind of notion of it's messy. It's not as organized as, like, functions set up and, you know, I need to build this, so I'm gonna bring three people from here and a back end and a front end and a designer, and I'm gonna bring it together. Like, you have to start forming it more on the fly. But you'll start seeing BD people building their own partnership programs because they can.

Why would they wait for engineering? You'll start seeing engineering focus on the hardest problems, and so on. Um, so I think a lot of it will be, in my mind, a lot more of a kind of less, what's it? Still structured, but less layers for sure. Uh, I think I saw recently a company, without naming names, moved from 2,500 roles and kind of role families to 60. It's a massive compression of the stack. And then you'll start seeing a lot more evolution of small teams tackling problems and evolving really quickly without kind of the need for. That's the best companies.

They will be very agile and adaptive. The worst companies will stay in their lane and do stuff the way they're used to do. [JENNI STEIGER] I think we have time for one more question. [HANK] Yeah. Thank you for being here. My name is Hank, MBA one. Kind of piggybacking on what you mentioned earlier, I have a question more related to reskilling and learning and development. From your perspective, what do you think, for instance, LinkedIn Learning has to do right to capture this

waves of needs in the labor market where a bunch of people need to learn AI? Or if there are other players that you find very promising. [TOMER COHEN] Uh, promising in the sense of what? [HANK] In the sense of they can do this very well, or perhaps this will just happen organically. [TOMER COHEN] They can teach or they can learn really well? [HANK] They can teach, like enable people to learn very well.

[TOMER COHEN] Yeah. I think this is one of the hardest problems in front of our economy right now. Like, 'cause there's already a massive skills gap. There's always been-you know, LinkedIn has this amazing bird's-eye view into the world's economy. We were seeing a long time, like, what's the skills needed, the roles needed, the functions needed for a very long time. And ever since I joined the company, the tagline has been, "There's a massive skills gap." And that skills gap has always been growing. Because change is like the fact that we're seeing this insane amount of change right now, but there's always been growth.

Like, if I give an example right now, if we look at the jobs on the rise, the most in-demand jobs right now, 70% of them were not on the list a year ago. Now, there's always been a lot of change in that list, but 70% is unprecedented for us. So the skills gap is keeps growing. I think this is where you can't just expect people and companies. Governments will have to start stepping in. There's going to have to be programs assigned to help people learn those skills.

Like AI fluency is going to be part of how you teach people to interact, just like you learn computers in school. But it's going to-- there is, there is going to be some period of, uh, instability there, actually. Something that actually really concerns me, that instability around those who have the skills and those who don't will become greater and greater and will lead to a bigger kind of bifurcation of capabilities. And then that needs to be closed. Now, partly schools will close it, but schools can't assign it to the whole population. So in my mind, governments are going to have to step in and play a bigger role in doing so. In terms of teaching, I think the best companies to learn this at

are the companies who are already working this way. because then, like, nobody's teaching you, it's the way we work. And that's the best way to learn. Like I always, you know, somebody joins the company, they usually have like a 100-day, you know, plan for how they're gonna start their role. But the best way to start their role is in a crisis. You just learn everything and it feels crazy and you don't sleep, but then you come out of it, you know everybody. Again, then you learn how to do it really well.

It will be in no management book, but I think the coming in from- through a crisis to a company, you just learn so much so quickly because you have to. And all those hundred-day plans usually, like, they don't work, they become boring plans, and you don't actually interact with them. So if you want to learn fast, the best places to learn it are at companies who are not trying to teach it, because this is already how they work. It's almost like when I was coming to LinkedIn, we were going through the mobile transformation. I came from mobile. The best way to learn mobile was to just work with people who worked on mobile.

It wasn't, say, "Teach me mobile," for somebody who was teaching mobile. It's just to learn on the job and do things. And that's why when you think about companies out there, and you think about where to go, assess them by those capabilities, not just by- Again, put aside the bonus, put aside the title. Focus on how much you're gonna learn in the first year. That would be my first-might even be my P0 for deciding where to go to. [JENNI STEIGER] Awesome. Thank you so much for joining us. [TOMER COHEN] Thanks, everybody.

(silent)