Lenny's Podcast: Product | Career | Growth
00:12:04 · Nikhyl Singhal
[00:12:04] S1: Yeah, I had this meetup last Thursday and it was a fun meetup. I have a group of about 125 heads of products, and we tend to gather once a month in San Francisco. During this meetup, we took an approach where lots of people were building things for their own companies or on their own to help themselves with productivity. We said, "Hey, why don't we do a show and tell? Why don't we just have you, if you're a head of product, just show things that are interesting, that are worth explaining." We had a startup go up and show what they're doing, and then we had some insights companies and some late stage companies. It was a few things that were really palpable. One was, there was so much joy in the audience that people were like, "Oh, I've been building this thing and let me show you." Everyone had their laptop and they were one-upping each other and they were like, "Your chief of staff does that. My chief of staff does this thing." That builds on my point that people are having fun being hands-on. But the second thing is that the way they talk about how product decisions are made, the way they talked about how prioritization is determined, the way they talked about how information moves within the company, looked like a completely foreign animal from anything that I had experienced when I was working a few years back. I got up on stage and I said, "If you think about how we work right now and how you folks are all talking about using agents in your enterprise, how you're using chief of staff apps to essentially drive more productivity. How all of you are essentially spending all your time just focusing on judgment and you're spending all your time taking anything that can be obsoleted and writing software around it." I think none of that was even in the language, even in the vernacular 12 months ago. Now, what is it going to look like in 12 months? Everyone had a pause and they're like, "Yeah, it was impossible to anticipate what this conversation would look like 12 months ago." It's pretty hard to understand where things are going. But I would say, I think maybe I'll answer your question around where things are going in the next couple years based on how I think companies are going to change and how people are going to change. I think on the company side, I think that, and then maybe I'll slant it more towards product for a second.
00:14:44 · Nikhyl Singhal, Lenny Rachitsky
Product leaders will increasingly get paid and be asked to drive judgment, and then be the tip of the sphere on trying to essentially obsolete everything else through software, through AI, through agents. This is partly why there's a lot of excitement, because just to be very honest with you and you were a PM and I was a PM, most of the stuff that you didn't like doing are the things that AI is really taking a real crack at. I think that's super interesting. I think that there's a ton of that change that's going to happen. Some companies are doing it now, some companies will do it, but within two years I think most of these companies will obsolete all the mechanical parts of building product. I think that there will be 10 to a hundred times more changes that will be presented to products than ever before, because now the cost of testing something, the cost of changing is going to be much, much lower. When things are changing that rapidly, that judgment piece becomes pretty much paramount. [00:15:52] S0: Just to make sure we understand, when you say judgment, what should people be thinking of when you talk about judgment?
00:15:57 · Nikhyl Singhal
[00:15:57] S1: I think that it's evaluating whether the thing that we're changing is a good or bad thing. I think that it's also evaluating whether we should change the product in one way versus another. You can't build a hundred custom versions of the same product. It affects your brand, it affects the maintainability, et cetera. When customers are asking for things, when you're trying to think through how do you build something that's sustainable, differentiated, that's judgment. Evaluating whether it's successfully met that criteria and whether it's worth building and worth releasing, it's almost like the system skill that's existed from the beginning of the internet. It's, "Hey, it's not about the feature, it's around the system that we're putting together, the platform, if you will, that's enabling capability." That's the judgment of, boy, there's all these changes coming in. They're going to happen more frequently. As an aside, because of all these changes, I think that in two years, I think there won't be any more bad software. This is maybe more of a wish and a dream than a prediction, but I think that if you count in the week, in your given week, how much bad software you come back with. I have a house and it has 15 different apps that run to control everything from the shades, to the air conditioner, to the garage door. Almost every one of those apps are afforded. They don't work particularly well. They never get monitored. Things break, no one ever checks it. All that stuff's going to get fixed because someone's going to basically sit down and tell Claude to fix it, and it's just going to fix it. It's going to fix it. It's going to be more secure, et cetera. I think that that's changing, and so people will have a lot less tolerance for bad software.
00:17:47 · Lenny Rachitsky, Nikhyl Singhal
[00:17:47] S0: Along those lines, just real quick at a quick aside to your side, the prompt I constantly find myself using with Codex and Claude is just how can we make the product experience better? How can we make this better? You just ask that and it's like, "Here's 10 ways we can make it better." It's like, "Wow, these are really good ideas." Then cool, do the first seven. [00:18:06] S1: Exactly. There's going to be some super skill that's going to run against every piece of software that's in the App Store, and it's going to basically go in and it's going to fix them all. Then they're going to get released and they're going to be like, "Oh, it's just a significantly more consistent, better, less buggy experience and more maintainable." That's why there's a lot of optimism because these changes can be done pretty automatically. [00:18:29] S0: Yeah. It's easy to, just to finish that loop real quick, so many of these apps are built by engineers that are not the best engineers. They're engineers that companies hire that don't really care about their software. They're just like, "This is just a side thing building this thing that we have just app. We just need to build an app. Let's find someone to build it." They're not like product first companies. Now, all of these companies have access to the most skilled software engineer, Claude Code, Codex and all these other tools. I totally see what you're saying. Everyone has the best engineers available now, and it's just English to ask them to build the better thing or build a better app. That's a really interesting point.
00:19:07 · Nikhyl Singhal
[00:19:07] S1: It still boggles my mind. Everyone's like, "Oh, what's the cutting edge and all of that." I think there's still, I don't know if this is true even today, but very recently there were more lines of COBOL than I think any other language out there. Mainframe sales continue to do pretty well and they're unusually large line item for companies like IBM. Part of the reason why these systems... I think I spent an hour and a half trying to get my MileagePlus account on United to work correctly with my daughter's phone and all this other stuff, partly because these systems are so complicated, but they're built in mainframes. A lot of the engineers, and this is going to sound morbid, but I think it's true, a lot of the engineers are dead. They literally are passed away. They were written 10, 20 years ago and going in and touching that code is probably the last thing anyone wants to do, so nobody goes in there. It's all that side. Now, we can change that. I'm very excited about what it means to go in and improve things that people take for granted. That's a big one. That's a bit of an aside. I think that the other thing that I think is worth signaling is, I think that the way the skills that you need to be effective in today's world is an acceleration of what we were talking about last episode. We were talking about ICs, we were talking about hands-on, we're talking about opinion, even back then, before Claude and ChatGPT and others came out. I think what we're now seeing is companies are looking very, very carefully at their staffing and they're asking themselves, one, do we need this many people? Did we overhire? Part of the reason is they're looking, "Hey, maybe we doubled our staff in the last five years." Lots of companies that are public have done that. "Did we get twice as much for them?" It was funny, when I was at Google, now over 10, 15 years ago, we used to ask ourselves, I was on an ancillary team. I wasn't in the search or the ads team. We used to ask ourselves, "How many people are really needed for Google to hit their numbers?" They were like, at that time, 20,000, 30,000, 40,000 people. If you ask someone who wasn't in tech, you're like, "Oh, probably 90% of those folks, obviously." The answer's probably closer to 9%. Probably you need 500 people to keep the lights on and to build that business.
00:21:52 · Nikhyl Singhal
You don't need 25,000, 30,000, 40,000 people. There's a huge amount of overhead where companies hire not just because they want to have bureaucrats, because they want to expand, they want to try new things. But I think that there's a judgment day that's come back where companies are like, "Look, we aren't getting as much for the staff that we grew in the last five years. This AI thing requires a totally different skillset." The combination is going to mean, this year I predict in the next 12 to 24 months, we're going to see massive shedding of staffs and then massive rehiring. But you might see a company shed 30,000 and hire 8,000, but the 8,000 people they're going to hire are going to all be AI first. The 30,000 they're going to let go of are going to be in combination of, we didn't get as much for those folks that we needed and we want to set the destination differently with a much lighter payload, and that's dark. That's going to make this year and the next year going to be pretty, pretty challenging. I'll transition a little bit to my thoughts on people, but I'll pause for a second to see if this resonates with you.