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

Keith Rabois hasn't used a computer in over ten years: he does everything from his iPad, phone, or watch. But that's not the most shocking thing. Keith says that in the future, the role of the product manager, the “PM,” will no longer make sense. Companies will be made up of people who can answer a single question: What are we building, and why? There is no longer any need for someone to put together one-year roadmaps, because AI is changing everything too quickly. Today, the capabilities of AI models are evolving at such a rapid pace that what you couldn't do in November is already commonplace in March. So, the winners will be those who adapt almost in real time, those who are the first to spot a new opportunity and seize it immediately. The real skill will no longer be managing processes, but figuring out what to do—and having the courage to change course every week. Keith puts it bluntly: “The role is more like that of a CEO. You have to decide what to build and why.” And this ability—knowing how to identify what really matters—is what sets true “barrels” apart from the rest of the team. “Barrels” is the term Keith uses for people who can carry out an initiative on their own, from idea to result. At PayPal, one of the most talented companies in history, out of 254 people, only 12–17 were barrels: they are the ones who truly make a difference. Everyone else is “ammunition” – useful, of course, but without the barrels, nothing gets done. Hiring more people without increasing the number of barrels only serves to create collaboration and coordination overhead: the classic case of spending more and producing less. One anecdote makes it all very clear: When Keith was at Square, he wanted cold smoothies delivered at 9 p.m. for the engineers who were working late. After a thousand failed attempts by the office team, it was a mere intern, Taylor Francis, who solved the problem on his second day. From that moment on, Keith knew he had found a barrel: someone you can entrust with a task and be sure it will get done. Here's the difference: a barrel sees an obstacle, finds resources, motivates others, realigns everything, and delivers the result, even if they don't yet know how. Everything else is secondary. But how do you find this kind of talent? Keith is ruthless: the ability to assess people is everything. He himself was mediocre at it at first, until he learned to poach the best talent from other internal teams. The trick? You don't need to interview a thousand strangers: it's better to focus on people you really know and, when you can't, learn to conduct reference checks without mercy. For example, Tony Xu of DoorDash conducts twenty reference checks for every senior hire. Another technique: Ask candidates, “If you were the CEO here, what would you do differently?” Not to get trivial criticism, but to find out if they have a strategic vision, if they can see what is really needed to create value. Another point that Keith challenges: in his view, talking to customers is useful almost exclusively in large enterprise companies. For everything else, it's downright harmful. “Customers don't know what they want, and they often give you completely misleading answers.” It's better to trust your own instincts, your own deep intuitions. It’s like writing a song or creating a work of art: if you follow feedback too much, you end up producing only derivative, soulless work. And here comes an even harsher truth: in ultra-high-performance organizations, “psychological safety” does not exist. “High-performance machines don't seek psychological safety. They want to win.” According to Keith, the CEO's job is to push harder precisely when everything is going well. The better the team is doing, the greater the risk of complacency. And true talents, like world-class athletes, are those who get bored if there is no challenge, if everything goes smoothly. When the pace slows down, morale drops. Another hallmark of winning companies, in his view, is speed of execution. When he was on the board at Square, an investor told him, “I haven't seen this pace since PayPal. You identify a problem, and a month later, you've already implemented the solution.” And this ties back to the hiring philosophy: the teams that perform best often don’t hire super-senior talent from outside the company, but instead develop people from within. An example? At Ramp, the CMO and the new Head of Product were both former Chiefs of Staff. You create a talent factory where people learn on the job. And what about AI? Keith notes that today, at the best companies, the CMO is often the biggest consumer of tokens: these are people who, thanks to AI, no longer have to wait for other teams, but instead experiment, analyze, and launch campaigns on their own—intellectual curiosity becomes the real competitive advantage. What about the future of roles? Those who can combine business acumen, technical skills, and the ability to adapt quickly will have a huge advantage. In this landscape, design and code are merging: the real “alpha” now lies in telling a good story, cutting through the noise, and finding the narrative that resonates. And if you want to understand human beings, Keith says, forget about market research: “Everything you need to know has already been written by Shakespeare.” The phrase to remember is this: The team you build is the company you build—everything else is just a consequence. If this perspective on talent and speed has made you rethink how you approach work, on Lara Notes you can press I'm In: it's not a like, but a way of saying that this idea is now part of you. And if, in a few days, you find yourself telling someone the story of the intern who solved the smoothie problem or why talking to customers can be detrimental, on Lara Notes you can tag the people who were with you using Shared Offline: It's your way of saying that conversation really mattered. That was Lenny’s Podcast with Keith Rabois: you saved over seventy minutes compared to the original episode.