Hard truths about building in the AI era | Keith Rabois (Khosla Ventures)

The idea of a PM makes no sense in the future. The skill is more like being a CEO now, which is what are we building and why?

AI is going to radically reorient lots of people's careers, maybe including mine. What I've noticed in some of the best organizations is the number one consumer of tokens is the CMO. They don't need to rely upon deputies and deputies and deputies to get actual work product.

I hate talking to customers. I refuse to allow colleagues of mine to talk to customers.

High performance machines don't have psychological safety. They're about winning.

If a founder shows the ability early in his or her career to assess talent ruthlessly and accurately, he or she can go very far with no other abilities whatsoever. It feels like it's never been harder to attract the best talent. Really talented people, when things are going well, they're not happy. The morale actually does go down when people are skating. The single role for the CEO is offsetting that complacency. The better you're doing, the more the CEO should push.

It's a pleasure to be with you.

Yeah. So, when I started working at Square, Jack Dorsey was running the company off an iPad. And so, I immediately converted in September 2010. And haven't looked back. I haven't touched a computer since September 2010. Everything I do in my life is either done from my phone, my watch, or my iPad.

Yeah. Try to be ahead of the curve. Jack's very good at being ahead of the curve. If you just watch what Jack's doing and follow, you're in pretty good shape in technology.

Yeah, partially distractions, partially just the flexibility. Like taking an iPad with you anywhere is just super easy. You know, some laptops have improved since then, but like the screen flexibility, angles, but like just the weight. Like I carry my iPad with me everywhere. So, there's no reason, there's nothing you can't perform unless maybe if you're doing heavy-duty engineering, which obviously has not been my forte in life, although I may have to start. There's no reason to use a more powerful, heavier-weight, less flexible machine.

See if you got my Apple. As long as it's an Apple product, it works.

Well, let's start with first principles. The most important lesson I learned when I was working at Square from my board was Vinod Khosla was on my board and he said the team you build is the company you build. And that adage is the most important thing when you're creating a startup. People get distracted with the market, with customers, with a product, with technology. Ultimately, it's a team. If you have the right people, everything else will be easy. And if you have the wrong people, everything else is going to be difficult. So I actually learned this, Vinod distilled it, but I actually learned this back in my PayPal days. So, you know, in the early 2000s, why was PayPal so successful? Why were there such subsequent generations of successful, interesting companies for, you know, 25 years now? It's because Peter Thiel and Max Levchin marshaled an incredible density of talent. So it allowed PayPal to succeed where possibly we wouldn't have. And these people went on with interesting ideas, ambition, and talent to build epic companies, you know, in all kinds of verticals. So from day one, I've always been, day one of my technical career, technology career, I've always been focused on the importance of critical density of talent. How do you identify, retain, and promote people with talent? Back in the PayPal days, when I first started my career in technology, I actually truthfully was not very good at this. Fortunately, Peter and Max were. So, Max basically hired all the technical talent in the organization. Peter pretty much hired everybody else. They used their network primarily. So, it was very difficult to get a job at PayPal unless you had a first-degree or second-degree connection to the engineering team or to Peter through Stanford, which is a different type of recruiting model. It works really well if you have a strong network. I wouldn't recommend it for everybody, but if you have a network that has, you know, unique talent, there's no substitute. Interviews are not a great substitute regardless of how strong you are at interviewing. But when I started my career, I was mediocre at, let's say, hiring people. Probably 50/50. You know, some good people, some mediocre people. That doesn't allow you to scale a team with an unfair advantage. But what I learned to do is steal people from other people's organizations within PayPal, very successfully. So I got feedback from David Sacks, who was the COO at the time, that I wasn't going to get promoted again until I could demonstrate leverage, like leadership leverage, which he had an equation for of 1 plus 1 has to equal 3 or more. So for every incremental person you hire, you have to show that you produce disproportionate returns, nonlinear returns. And because I wasn't hiring that well, I wasn't really succeeding at that leverage. So when I went around, took the feedback into account, I said, 'Hm, okay, I really want to get promoted. What do I do?' So I found people within the organization that I felt had talent that were not being leveraged to the highest potential and ambition, and I recruited them to my team, and that was very successful.

And then I did get promoted actually fairly quickly because these people actually were up to speed. They were able to run really fast and we produced a lot of really important results for the company. The lesson though that I took away was, well, that's great because what it showed to me is I actually could identify talent. I just couldn't identify strangers with talent. So people that were in the building that I have lunch with or dinner with or go for a run around the PayPal campus with, I was accurate at diagnosing their abilities. I just couldn't do it, you know, in a 20-minute, 30-minute, 45-minute interview. So the first hand is just double down on people I know. That doesn't scale perfectly, but learn to be excellent at it. If I had context assessing people's abilities, superpowers, and, you know, weaknesses, then over the next X years tried to identify a different technique for identifying, assessing random people. Because ultimately, if you're going to build an organization, ultimately if you're going to be a VC, you can't just invest in people you already know. So that took some years and required some training anyway. I think you can teach some of this to a founder, but one advantage a founder has that's going to thrive is if a founder shows the ability early in his or her career to assess talent ruthlessly and accurately, he or she can go very far with no other abilities whatsoever. Hiring is like a muscle. You need to exercise it. You need to try, learn what worked, what didn't, what could you have known, what did you miss, why? And, you know, riff on that and try to get better at it. There are tactics you can learn. I think the tactics work pretty well within the middle of a bell curve distribution of moving yourself 10, 20 degrees within that bell curve. I don't think the tactics can really teach you how to identify, call it like top 10 basis points, top 50 basis points of talent. There you have to deviate from the norm. And I think that's actually true of most lessons in life. If you're going to be extraordinary at any skill, you can't follow a playbook. Otherwise, by definition, lots of other people would be, you know, the top 10 basis points. But you can get a lot better by learning techniques. So, for example, let's share a couple. One thing I think you can learn to do is be excellent at references. It doesn't work for hiring people right out of college or something because the reference context is going to be a little off. But I think without getting better at interviewing and assessing, if you just learn to extract the right information from ruthless referencing. So for example, ruthless referencing to me means Tony at DoorDash does 20 references on every single senior hire. 20.

I bet you he's pretty good. I bet he's been pretty accurate, too. So I think you can learn that. That's like a skill that's teachable, that being absolutely, you know, incredibly dedicated to your craft, you can just get better at. Back in the day, there's an investor at Greylock, who's on my board at LinkedIn, David Z, very successful investor, notable for both LinkedIn and Facebook investments. He used to teach at Greylock, you couldn't stop referencing a founder until you hit a negative reference. So you would know you had exhausted the reference when you finally hit a negative reference. And so I think there are tactics there in, you know, muscle building. How do you get the right information from the right people? How do you frame the questions, etc., that will lead you in the right direction. Now you have to be careful. Like, let's say I'll give you an example where this can go wrong. I've been a longtime investor from the seed round of a company called Fair, founded by two of my colleagues at Square. Max R and Jeff Cverson worked for me at Square, and then the two other co-founders also worked at Square. When people were reference checking Max, often most VCs asked the wrong question, which was, was Max a good employee? The answer to that is very mixed, and so some venture capitalists, including some very good ones, were nervous about investing in Fair. If they framed the question slightly differently, which is, is Max capable of being a world-class entrepreneur, the answer was yes. So again, it's like a tactic you have to understand, like what exactly am I trying to extract? Same person, wrong question, wrong result. And many people passed on Fair and they regret it. And these are actually quite talented investors, they just didn't frame the question correctly when they were calling someone like Jack Dorsey up for the reference.

When I interview candidates for senior people in leadership positions, I always ask them, you know, look at whatever company they're at and say, 'If you were CEO, what would you have done differently?' And you get a feel for their strategic mindset of, you know, can they drive value creation? Because almost by definition they've come from a company that's had some traction, success. So can they end it? So for, in your case, I would have asked you, you know, if you were CEO of Airbnb, what would you have done differently? And you learn a lot from that question. On reference specifically, I think a general arc that's pretty good is asking the person what would lead to this person being most successful. And if something were not to work out, what would be the primary root cause that you can identify of something going wrong? I think generally probing on those two arcs leads to a lot of insight.

For the candidate. That's for the candidate specifically.

Because it's not like I don't want them to criticize Airbnb. I don't think that's that productive. But you can tell how much of the current business model have they absorbed, how much they understand tradeoffs, and then can they create an unfair advantage? You know, can they have insights into afterburners? And then I have a follow-up question which is usually gold, which is, let's say I ask you this question about Airbnb and you give me this great answer, I'm like, well, why didn't you, why weren't you able to persuade Brian to do it?

So I've read some research on the topic, and if you ask yourself 30 days after any hire, would you make the same decision? That 30-day loop is pretty useful, and it basically is as accurate as measuring it a year or two years out. So you got a pretty tight feedback loop, and you can ask the entire hiring team. So I think that is like just a technique that every company should use.

You definitely need ammunition, and it depends on what kind of project. There are types of projects where an individual barrel may be able to succeed with very limited or no ammunition. Sometimes it may be a designer, an engineering team, a PM, a data analyst, blah blah blah. It depends on what the project is, what the problem you're trying to solve is, what's the proper amount of ammunition. But once you think about the ratios of ammunition to the problem, you can be much more constructive and deliberate and intentional about the team construction.

Basically, we're going up that there's a hill over there. That's the hill. Get us over that hill. And one way or the other, they will motivate people if they need to. They will accumulate resources if they need to. They will measure what they need to. And they're going to get your company across that hill. That's a barrel. Anything less than that is not a barrel.

Yeah, it's kind of the collection of all the things together.

Co-founder wants an outcome, and come hell or high water, this person is going to deliver that outcome. Then the outcome can be, you know, fairly narrow and not that difficult in the beginning, and then you expand the scope, you know, the complexity, the difficulty that you basically entrust to your barrels. And sometimes they have no line of sight of how to solve it when you start. Sometimes you have a preliminary idea. So it ranges, but ultimately it's that skill of, I'm going to take this off your plate. You can fire and forget, and this is going to happen. And it's not going to happen. I am going to come back to you proactively with the issues I'm confronting, what I've already tried, the diagnosis of the root causes, and ask for your help with sufficient time for you to intervene and try to brainstorm with me to get us to the right answer.

Yeah, I think agency is accurate. The problem I have with terms like agency is it's like a little bit like strategy because in one ear a lot of people went and out the other and they don't really process the meaning.

You know, I talked to my YC lecture in 2014 about how to operate. They can be as simple as the now somewhat famous in technology smoothie test, which is, you know, we used to have engineers work pretty hard at Square and pretty late. And I always wanted them to have like food so they wouldn't be famished, they wouldn't be distracted. And I didn't really want them to eat like junk food because I actually think junk food is bad for you, bad for your brain, etc. So I settled on delivering and really wanted to provide like a 9:00 p.m. like cold smoothies. And we had at that time a pretty substantial team at Square, office team, EAS, you know, this was not a lean mean organization. And so I tried through the office team, EAS, and nothing. We never got healthy, delicious, and cold smoothies delivered at 9:00 p.m. Just kept that one. It's getting frustrating because if the smoothies aren't cold, then no one's going to eat them. If they don't arrive at 9:00, no one can really bank on, you know, the refresh rate, the refreshments. Everything went wrong. And then I had this intern named Taylor Francis. And I was explaining just my frustration. It was like his second day, and he's like, 'I'll solve it.' And I was like, 'Okay, kid, you know, good luck with that.' Like I was like, 'Sure, keep trying.' Anyway, day goes by, 9:00 arrives, and lo and behold, smoothies show up at 9:00 p.m. delivered on the standing desk table where the engineers would congregate. I sample them. They're cold. They taste great. And I'm like, 'Oh my god, I found a barrel.' And I later gave him almost everything to do.

The standard stuff is still true. Mission, selling the vision and mission is indispensable. Most people who have proven talent anyway, at least in the current world, are going to attract offers from multiple, you know, opportunities. And so you've got to convince them that your opportunity is very special. I think one way to do that that's a little bit more nuanced is convince them that their particular skill overlaps with the critical blockers to the current company. Meaning they're betting on themselves. So for example, if they are superb at let's say marketing, if the biggest blocker in the company's current success is not technology, not the product, but we believe it's marketing, it's really easy to go to a world-class marketing person and say, not only is this great company building something really cool and interesting that you'll be proud of, but your particular ability is very unique and differentiated and you can solve this. This is actually how I wound up at Square back in 2010. I was actually, I'd just been acquired into Google and was planning on being a VC actually next after I was kind of like vesting whenever Google is going to compensate us. And then the investors in Square called me up and they said, 'Hey, we've been looking for almost a year now for someone who knows something about financial services yet is still entrepreneurial.' And they were like, 'Hey, there's only three of these.' At the time, 200 was like, 'There's only really two or three of you in the world, you know, so would you be interested?' And I said, 'Well, maybe.' But that was the argument to me that made me leave Google early, like after two weeks, and infuriate everybody and bypass, you know, venture for another three years, which had been my plan, was because they made the argument that, hey, I was one of three people in the world that could actually do this job. So there's like a, I don't know if it's ego, but it's also just like impact.

So I think that can be extremely helpful. You know, my more important arc in this is I think you have to build a company on undiscovered talent. Like I don't think you really want to compete for the people that everybody else wants, right? And you know, I learned this at PayPal. Peter taught me this literally the first day and first week of my job at PayPal that the way to build a company, we were jogging around the Stanford campus, is we've got to find these undiscovered talent. That's the only way to scale organization against these large incumbents with infinite money, etc. And you know, I've been on that crusade for 25 years. For those who are interested, you can link to it. I gave a speech at Ramp, you know, how to hire, but talks in detail. I also recommend Eric, CEO of Ramp's speech, which is fairly similar. Both videos are online.

Well, they're also not maybe the people you want. So there's adverse selection, but you know, it's like a salary cap. Most sports these days have salary caps. And when you're a startup, not only do you have a salary cap, you probably have one-tenth the salary cap of the people you're competing with. So you've got to figure out how to leverage, you know, less assets to more success.

I think it's basically isolating why other people aren't going to process them correctly. Like most recruiting at large organizations becomes sort of a homogeneous function. And so if you understand why this person is going to get thrown into this black box kind of thing and not get processed accurately, it's pretty easy. So I always think about, you know, let's say this person was interviewing at Meta or Google or Block these days or Coinbase. What are they going to miss and then why? And then that leads to, oh, perfect. So sometimes it's just lack of information. Like one of the reasons why, you know, sometimes it's controversial to say this, but one of the reasons why the net impact of my hire undiscovered talent is you wind up skewing younger. It's not because you need young people. It's that younger people have by definition less data. It's like a, you know, we use credit scoring, FICO scores. It's the same thing for employment. By the time you're over 30 some odd years, there's so many data points about you that this black box machine is usually going to process you like many other people. If there's no data points, it's very hard for a black box machine that does homogeneous evaluation to evaluate you. So there is alpha, so to speak, by definition for people who have like no data points.

Yeah, absolutely.

Yeah, I think that's true. I mean, I think look, ultimately, I'll channel someone else's feedback, but it's the same vein. So a friend of mine who's a CEO once asked Mike Moore, it's like, what's the most common denominator of the best CEOs ever? And he said, it's the relentless application of force. Quote, I think that's the job of the CEO. People eventually get comfortable, complacent. The more success you have, the more complacent the organization tends to get. And the single role for the CEO is offsetting that complacency. So for the point, the more success you have, the better you're doing, the more complacency naturally kicks in. And unless you've erected a network effect, you do not want to get complacent. And even then debate whether you should, but like fundamentally most businesses are not network effect businesses. They are not going to run on, you know, on their own for a long time. So I think that's one insight is the better you're doing, the more the CEO should push. Secondly, it's a little bit like sports when you're growing up. People when they're winning take advantage of feedback better than when they're losing usually. Like so for example, you know, now what I do is mostly VC, mostly a board member, mostly like a consigliere to founder. And when a company's struggling, maybe what's less intuitive, and you may have picked up on this in your research and, you know, interviews of people who know me, when a company's struggling, I'm actually usually a very non-critical and more like a coach and supporter because the company, the founder knows they're struggling. Being critical doesn't really help them solve their problems. That's when being supportive can actually somewhat counterintuitively be more important. But when the company's thriving, it's really important to be critical and isolating things that will eventually be problems while everybody in the company is really happy and borderline complacent. So you kind of want to be the opposite as a default. And that's like a really good sports coach. When you're winning is when to polish everything and really master the details. When you're losing, you definitely also have to be exciting people and embracing the future and selling the future.

Yeah, if you're doing great. Well, also you have to remember like I remember giving a speech once in Square is like you get to a certain threshold creates inflection momentum gets you a certain, you know, valuation and all these like attributes, but it's kind of like winning a Super Bowl. You get the last year was great. Last four quarters were wonderful. It's like winning the Super Bowl. You got to come back next year and start your record zero zero again. And you got to remember that actually venture is like that. You know, I'm only as good as my last investment. I've had like 13 years or whatever of pretty damn good investments. But like truth, I have to wake up every day and find some undiscovered founder that's going to change the world. And if I don't do that, it doesn't matter what I've done the last 13 years. And a company's kind of like that. The company can skate on autopilot for a while. Venture, you really can't ever skate.

Brian and I are usually in sync a lot. There's a really good interview where I interviewed him also at the same conference of how to hire when he talks mostly about founder mode, but I generally subscribed to virtually all of Brian's views. He even taught me some of these things himself. But the more important point I think you identified, which is very subtle, is really talented people are like superb athletes. And when things are going well and people are really kind of coasting, they're not happy. They have an internal clock tempo. They just want to create things and create value and drive, drive. And like, you know, like the morale actually does go down for the best people in the world when people are skating.

Well, I do think AI is going to radically reorient, you know, lots of people's careers, maybe including mine. So I think that's actually true. And I think the way to thrive in a rapidly emerging technology world is to be intellectually curious. So for example, you know, I'm a business person historically. You know, I did actually code when I was really young, but like basically professionally just a business person. What I've noticed in some of the best organizations is the number one consumer of tokens is the CMO.

And so they're like, 'Wow, there's all these cool things I can do now with my hands. Either I had to rely on other teams or never got access the way I wanted and blah blah blah,' and they just do it. This is actually true at Open Door. It's true at another great company that I'm on the board of that's incredible. And so I think you can be intellectually curious and future-proof yourself more than just, yes, you can work harder. I'm a big subscriber to like no days off and working all the time and all that stuff, but fundamentally the intellectual curiosity is able to learn new things and that is how you embrace the future.

Both massive, you know, awesome companies with lots of engineers. And I think that's very encouraging. You know, for the executive particularly, like this is definitely like the best executive in the company.

Well, sometimes it's like more like what we would have thought of analytics.

It's just like they don't need to rely upon deputies and deputies and deputies to get actual work product. And so they're just like shipping things and like or shipping, you know, drafts of things or giving the CEO insights into things themselves.

Well, I saw this podcast or listened to this podcast that Peter Fader did, and he convinced me that the idea of a PM makes no sense. Basically, in the future, if you think about decomposing the logic, is what does a PM usually do? They take these inputs from customers. They create this sequential roadmap that's well organized over the next year. Blah blah blah blah blah. That world is like ridiculous. Like right now the capabilities of foundation models or companies like Lovable and, you know, things like that are just so improving at such a rapid rate that it makes no sense to have a year-long roadmap and they're just like incoherent. There are things that were impossible to do in November that are actually pretty easy to do right now in March. And so I think you need to build an organization that's incredibly adept at, you know, people say nimble and all this stuff, but incredibly adept at changing the roadmap almost on the fly. And I think intermediaries like conventional PMs don't make a lot of sense versus being prepared intellectually, embracing, exploiting, noticing. So someone needs to notice that, oh wow, we can actually do this, but then exploiting it this week is the future of a very high-growth stellar startup will notice that something's now possible this week and create new features and new value for customers next week.

I agree with that. I actually think like whether you talk about someone who used to be a PM or someone used to be a designer or called an engineer, the skill is more like being a CEO now, which is what are we building and why?

And to be a successful engineer, that trait's critical to be a successful designer because the tools and the ability to actually create the thing, an object, is going to be easier and easier. But the art is knowing what to build. You know, another competitor of mine, Alfred Lin at Sequoia, likes to talk about being a chef. When you're a chef at like a prime restaurant, you're not actually cooking the dish. You're sampling, you know, your colleagues and, you know, editing their work a little bit. But fundamentally, it's half a commercial role. Being a chef in a famous restaurant is what's our value proposition, how do we differentiate ourselves, how do we brand ourselves, what's our segment, what's our pricing, you know, etc. What's our location even? That's what makes, you know, a famous chef. It's not they're literally cooking the dish all the time.

Business acumen. It's basically a business acumen.

Exactly. I understand the company's business equation, where we're trying to go and what the inputs and connection outputs are.

On my own...

If they were exceptional, I think that's right. But I think a lot of the best engineers I've worked with have commercial instincts. Like Max Levchin has this on steroids. Jeremy Stoppelman, you know, who's worked with me very closely at PayPal before he started Yelp and got promoted to be engineering director and vice president of PayPal, had commercial instincts, you know, back when he was an individual contributor. So I think there are great engineers who are technically proficient that have always understood the business building.

It's going to put a premium. I think this will, at the age of AI, will put an incredible premium on that because they're not going to need a large team. You're not going to be marshaling the forces.

I think design is going to be transformed as well. The tools are going to make it much easier to produce designs, but the taste and the judgment of what to build, what the experience should be, that's still going to be a human skill. And I think the best designers are going to be the ones who can think about the business, think about the customer, and then use the tools to produce things much faster than they could before.

Well, you know, it's interesting. I think design and code are merging, and it's not clear to me who triumphs — whether it's code becoming design or design just translating automatically into code. I bet on both in some ways, but I think they're merging in a way where they're not separate things anymore.

What I'm seeing is something really interesting happening with design. On the one hand, I just did some analysis on the job market for design and it's basically plateaued in terms of the number of open design roles over the past three years — it just hasn't gone anywhere. It's flat. We had the head of Claude design, Jenny Wen, on the podcast, and she had this kind of insight that in the design process, there's no time for the traditional design process. Engineers are shipping 17 things a day. There's no time to sit there and help mock and prototype and all these things. Well, let me give you a couple of concrete examples. So, at Shopify, the way they've developed — they've been doing this for over two years now — so this may seem normal, but they have not let PMs provide PowerPoint or Keynote presentations on product for like two years. Every presentation on product has to be a workable demo. And they just expect the PMs to create the products, and the execs just refuse to look at static, you know, 'we're going to have this feature.' No, I want everything working. This is for two years. So, I think that, you know, again, like everything's just merging together.

I do agree with that. I do agree with that, actually. I think the alpha is in design. Just like in marketing, it's not the tools, it's not the channels, not the metrics — it's the storytelling. It's how do you cut through the clutter in the snappiest, most compelling possible way. There's kind of an NP problem — there's so many different words you can use to express the same concept. But the person who can say this is the way to frame it, you know, the proverbial 'a thousand songs in your pocket' is worth like all the tools in the world.

Cutting through the clutter. I mean, that's always been critical. One of the times when people are pitching me as an investor, it's one of the things I'm dialing into, you know, immediately — how the hell is this going to cut through the clutter? It's one of the reasons why I don't like to take customer feedback into account, because by definition, when you put something in front of a customer, that's not a proxy for the real world. In the real world, you have to cut through the clutter while they're going to their Barry's boot camp, while they're doing their job, while they're raising their kids, you know, etc., etc., while they're walking, while they're on the subway. Like that, you know, an isolated fake experiment doesn't give you actionable insights and often is directly misleading.