Peter Deng(Meta、Instagram、Uber 产品负责人)
Peter Deng
Quick Fire Opening Questions
Lenny Rachitsky: You built and led Facebook news feeds. You shipped the Messenger app as its own app. You launched ChatGPT Enterprise. What’s an important lesson you’ve learned about what it takes to succeed building something from idea to one to billions?
Peter Deng: You have to plan your chess moves out in advance. You have to really think before you act and build systems that were going to let you go sustainably faster.
Introducing the Guest
Lenny Rachitsky: What’s the most counterintuitive lesson you’ve learned?
Peter Deng: Sometimes your product actually doesn’t matter. At Uber, I learned this because, really, the price and the ETA at Uber was the product. Looking at it from a holistic perspective, we humans consume the entirety of the product. It’s not to say that you shouldn’t fix the bug, but it doesn’t have as much of an impact as something that is more important to people.
How AI Will Change Education
Lenny Rachitsky: What’s one specific thing you think will change in a big way with AI that people don’t think enough about?
Coding and Abstract Thinking
Peter Deng: Education is going to change. My son, he was nine at the time, built a custom GPT that you can type in any topic and it would give you a sentence that had every letter of the English alphabet. Isn’t that mind-blowing? I can already see his brain rewiring.
The Power of Language
Lenny Rachitsky: What’s one thing you look for in people you hire?
Peter Deng: In 6 months, if I’m telling you what to do, I’ve hired the wrong person. It helps me and the person operate on a different level where the goal is not, did you hit this OKR? The Meta goal becomes, are we calibrating enough? Are we actually getting into a spot where in 6 months you’re the one telling me what needs to be done?
Returning to Product Topics
Lenny Rachitsky: What’s something you’ve learned about what it takes to be a great product person?
Peter Deng: I think there are five different types of product managers. Number one is-
Counterintuitive Product Lessons
Lenny Rachitsky: Today my guest is Peter Deng. Peter is maybe the most under the radar impactful Product Leader that you have never heard of. I often say that the best product people are not the people on Twitter and LinkedIn sharing advice, but the people who don’t have time to do that because they’re too busy doing the work. Peter is the epitome of this. He was VP of product at OpenAI where he oversaw product design and engineering for ChatGPT and helped ship ChatGPT Enterprise, voice, memory, desktop, custom GPTs and more. He also oversaw and built their first growth team. He was the first Head of Product at Instagram where he worked closely with Mike and Kevin, and oversaw all product development, including on content sharing, ads, growth, even helped build out their design and user research functions. He was also a Head of the Rider product team at Uber where he oversaw everything in the Rider app, including big improvements to pickups and drop-offs at Uber Pool and airports.
He also helped the team launch new products including Uber Reserve, which is now approaching a $5 billion a year business. He also spent nearly 10 years at Facebook as their 4th ever Product Manager where he built and led the team behind the current Newsfeed product, the standalone Messenger app, also photos, and groups, and homepage, and profiles. He was also Chief Product Officer at Airtable where he helped the company systemize how they built products and transitioned to Enterprise. He also led product management at Oculus. These days he is General Partner at Felicis where he is able to bring everything he’s learned to more founders as an investor. He has never done a podcast before or shared any of these lessons or stories publicly. So, you are in for a real treat.
A huge thank you to Eric Antonow, Nick Turley, Lauren Motomati, Joanne Jain, and Sundeep Jain for contributing questions and topics. This conversation, if you enjoy this podcast, don’t forget to subscribe and follow it in your favorite podcasting app or YouTube. Also, if you become an annual subscriber of my newsletter, you get a year free of a bunch of amazing products including Bolt, Linear, Superhuman, Notion, Perplexity and Granola. Check it out at lennysnewsletter.com and click bundle. With that, I bring you Peter Deng. Many of you are building AI products, which is why I am very excited to chat with Brandon Foo, founder and CEO of Paragon. Hey Brandon.
Brandon Foo: Hey Lenny. Thanks for having me.
Lenny Rachitsky: So, integrations have become a big deal for AI products. Why is that?
Brandon Foo: Integrations are mission-critical for AI for two reasons: First, AI products need contacts from their customer’s business data such as Google Drive files, Slack messages or CRM records. Second, for AI products to automate work on behalf of users, AI agents need to be able to take action across these different third-party tools.
Lenny Rachitsky: So, where does Paragon fit into all this?
Brandon Foo: Well, these integrations are a pain to build, and that’s why Paragon provides an embedded platform that enables engineers to ship these product integrations in just days instead of months across every use case from RAG data ingestion to a Agentic actions.
Lenny Rachitsky: And I know from firsthand experience that maintenance is even harder than just building it for the first time.
Brandon Foo: Exactly. And we believe product teams should focus engineering efforts and competitive advantages, not integrations. That’s why companies like U.COM, AI21 and hundreds of others use Paragon to accelerate their integration strategy.
Lenny Rachitsky:
Pragmatic’s full-time instructors each bring over 25 years of hands-on leadership experience, teaching strategies proven to deliver real-world results. And it’s not just about what you learn, it’s also about who you learn it with. Completing a course connects you to an active community of over 40,000 product professionals. You’ll engage in meaningful conversations, collaborate with peers and mentors, and gain direct instructor access to refine your strategies and stay ahead of trends. Get 20% off with Code LENNY20 at pragmaticinstitute.com/lenny. Peter, thank you so much for being here and welcome to the podcast.
Product Experience vs. Business Logic
Peter Deng: Thank you. I’m so thrilled to be here, really honored. Looking forward to having a great time here.
Why Tech Breakthroughs Aren’t Necessary
Lenny Rachitsky: As we were preparing for this conversation, we were jamming on what we should focus on. There’s so much that we’re going to talk about. But something that you said was really interesting and I’m really excited to start with this, which is that, you’ve always felt that you haven’t been able to say all the things you really think and feel because you’ve been within corporations, PR people keeping you on message, and this is the first time that you feel free to share.
Peter Deng: First time.
Moats in the AI Era
Lenny Rachitsky: Okay, so first of all, just how does that feel? Second of all, tell us something that you’ve been wanting to share or that you can finally talk about.
Peter Deng: Well, it feels really good. So, let me ask… I love it that you’re starting with a spicy question here and let me share some more context behind it. I’m here to speak more freely, but it’s not really what you think. I’m not here to divulge any secrets from the companies. But naturally I’m kind of a storyteller, I’m kind of an introvert. So, this podcast, I feel like I have the ability to go deeper with you on any topic and kind of add the context. Because I think without some of the context, some of my spicy takes or whatnot might be taken out of context, and just not having the time pressure, not feeling like there’s some PR message I have to hit, is just really freeing. So, it feels awesome, really anything that is on your mind that you should find interesting to your listeners, I’m here for it and yeah, I’m excited.
Building a Data Flywheel
Lenny Rachitsky: Something I always tell guests, and I don’t want people to take this out of context also, but I always describe myself as a reverse journalist where I want the guests to be the best version of themselves. I never want to catch people off guard or just say something they never meant to say. So-
Peter Deng: That’s great.
Product Craft and Distribution Advantages
Lenny Rachitsky: … it’s a safe space. Okay. But still, is there anything that you want to share or that might be interesting to share that you’ve been wanting to share that you haven’t been able to? Is there anything along those lines?
Peter Deng: I mean, I always get this question around sort of, AGI, is it coming? Is it going to solve everything?
The Role of Product in AI Companies
Lenny Rachitsky: What have you seen?
From Zero to One to Scale
Peter Deng: I mean, it’s so interesting because when I was at OpenAI, it was around the time that people were really scared of AI and, “Oh, it’s going to get rid of humans or it’s going to do all these things.” But with every technology, I think everyone’s been just kind of taking some time to acclimate to it. And I think with AGI it’s a similar thing, which is it’s so far out that everyone’s like, “Well, what’s our world going to be like?” And the real answer is none of us really know. But in terms of solving problems, I think some people believe AGI is going to solve everything, but I don’t think so. AGI is just necessary but not sufficient. A lot of the value is still going to require a bunch of hustle from a lot of builders to really turn that new source of energy and channel it into something that we humans want to use that solves some of our problems. And that hustle is going to be required, that elbow grease is going to be required to really make AGI something useful.
Lenny Rachitsky: Your point is that people think AGI hits, all of a sudden all jobs are gone, AGI is doing everything. Because I think this is a optimistic message that things will be okay if AGI, basically AGI being, and I’m curious if you have a clear definition, but AGI being, AI being just basically as smart as humans-
Portfolio Thinking and Resource Allocation
Peter Deng: Look, I won’t-
Lenny Rachitsky: … generally.
Measuring Everything: Value of Growth Teams
Peter Deng: … claim to be an expert on this at all, but I think that with every technology that’s come out, we’ve been able to harness it and it takes a lot of harnessing. I think I’m going to use that word very deliberately. I’ll use something really basic. What seems obvious today is that, there was a time when databases were all the rage. It’s like, “Oh my goodness, you can store a bunch of data and you can query it really quickly and imagine all the possibilities.” And I think that a lot of amazing entrepreneurs and builders built some really great products on top of databases.
Taste and Craft: Balancing Growth and Quality
Lenny Rachitsky: That’s right.
Building an Avengers-Style Team
Peter Deng: In fact, that’s kind of the basis of all the stuff that we’re seeing today. And it seems so obvious today, but I don’t know, maybe in 10 years, 15 years when we look back, it’s like, “Of course it made sense that we have this super intelligent thinking machine.” But it requires the product builders to be able to go in there and say, “How do we channel this energy to make it something that we as humans love to use and want to use?”
Lenny Rachitsky: I love the optimism around this. It’s just like things will not go crazy once computers are as generally intelligent as humans.
Five Archetypes of Product Managers
Peter Deng: I think that’s exactly what I’m trying to say. And I think that again, every technology people have this fear. And I remember watching a documentary once and they were talking about how when the bicycle came out, people were like, “Oh my goodness, this is going to be the end of all things.” And again, it sounds silly today. Because you’re like, “bicycles, really?” But then if you put yourself in the context and the mindset of a previous generation, which the next generation will be looking back at this podcast in that previous generation, I think that again, I think optimistically, things are going to be okay, we’re going to adapt. And this was actually one of the things that I talked about with my friend Josh Constine at South by Southwest, is this idea that humans will always co-evolve with technology. And I think that that co-evolution is already happening.
If you take a look at, there was a lot of a fear of AI just when ChatGPT came out, but when you start to get familiar with it things, that kind of things change and then you are able to evolve from being fearful to familiar and to go all the way to having this mastery of this thing of like, “Oh my goodness, look at all the startups that are happening now. All the things that we can build. And just over 18 months.” I would say we look back and there’s been an attitude shift. And so I guess part of my optimism comes from, if you look back 18 months and you look forward 18 months, might it be the same thing for something that we’re chasing now?
Lenny Rachitsky: Well, let me follow this AI thread a little bit more and then we can move on to other things. I feel like every conversation, there’s a time to AI conversation and then it’s like, okay, there’s other things that also matter. So, let me ask you this, the question, what’s one specific thing you think will change in a big way with AI that people don’t think enough about?
More PM Archetypes
Peter Deng: I think education is going to change in a big way. And I think a lot about this because I’m involved in my kid’s school quite a bit, and that’s something I’ve done after I left OpenAI. And what’s fascinating to me is that watching my son who got to dog food, a bunch of the OpenAI stuff before it was public, I think I can safely say that, that seems okay. And when he was playing with ChatGPT and some of the latest models and he was nine at the time, I can already see his brain rewiring. He was starting to ask questions and he never heard the word prompt before, but he’s like, just this is how awesome the human mind is, because he was exposed to this technology at an early age, some things just are unlocked. And I think that you’re able to think differently. And I’ll give you a specific example of what I mean here.
He goes to Python class and he’s coding. Now, I don’t actually think he’s going to have to code when he grows up. I think that’s going to be a solved problem. But it’s a very valuable skill because I think learning to program is learning how to think in a structured way, in a systematic way. And he was prompting ChatGPT with some really crazy things that I never even thought of. And one of the things was, “Hey, ChatGPT, can you give me a sentence that has every letter in the alphabet along the theme of oceans or along the theme of space?”
And the reason this kind of blew my mind is because in traditional programming you couldn’t write that program. You can’t say in Python like, “Oh, write a function that goes and formulate.” I mean, it’s a really difficult function to write. But for him to be able to think of that prompt, which is really cool because he built a custom GPT that you can type in any topic and it would give you a sentence that had every letter of the English alphabet, kind of like the quick brown fox jumped over the lazy dog. Isn’t that mind-blowing?
At age nine he could think about that, whereas being at age nine, I was playing with Legos and maybe QBasic. And so this idea of how young human’s brains will evolve because of this new tool we have is going to change the way I think we’re going to do education. And I’ll be very honest, I’m not an expert in education, but I just thought a lot about it. And one thing I think is going to be really important in the future is being able to figure out how to ask the right questions. We humans are inherently inquisitive. But being inquisitive and turning that into the right questions to prompt or ask AI, which is going to be again, something that everyone’s going to have access to is going to be a differentiator for what kind of work can be done.
The analogy I’ll draw is, when the calculator was invented people didn’t stop doing Math, they just did higher level Math. And it frees the mind up to do other things and think more at a higher level of abstraction. And I think we got to prepare our kids on thinking about, “Well, how do you think at a higher level of abstraction?” And this has happened before. I think Google has made memory kind of obsolete. You don’t have to memorize facts anymore, you can just Google it. And the next phase will be something around, “Well code will just appear if you summon it.” So, what are the things that people will think about and the skills we have to develop that are at the next level of abstraction, that tap into our creativity, that tap into our curiosity? That’s going to be really interesting. So, I think education is going to change dramatically, just like how progressive education in the past switch from memorization of multiplication tables into something that’s a little bit more kind of higher level, higher level thinking. And I think that’s going to be one of those big areas.
Lenny Rachitsky: This makes me think about an NPR story I was just listening to where they were following professors using ChatGPT to create their curriculum. There was a lot of talk of students using ChatGPT, cheating, having ChatGPT write their essays. But teachers are using ChatGPT in a big way. And then students are raiding professors badly because they noticed they’re using ChatGPT for their curriculum. So, it’s kind of this arms race.
Founder-Market Fit
Peter Deng: But it’s also interesting because then that it goes further, it show further though. The whole system has to change. Because again, I still believe that human brains are inherently inquisitive and that we still need development in some way. But how that’s going to develop, I’m fascinated to watch how that plays out.
The Art of Hiring
Lenny Rachitsky: I want to get back to product, but first of all, I know something that you think a lot about along these lines. This came up in many conversations I had with folks that you worked with. Is your emphasis on the power and importance of language, being really good at thinking about the words you use both in writing and speaking. Just talk about how you think about that, just the importance and power of language as a leader.
Peter Deng: I remember taking this class that really stuck with me in college. It was called Language and Thought. And it was taught by Herbert Clark. And he had this thesis that kind of blew my mind, which is that, “Language actually affects the way you think.” That’s one of the parts of the thesis. And once I heard that and read that in his book and listened to the lecture, I couldn’t stop thinking about that because it just rang so true. I grew up speaking Chinese and I think that there’s a lot of things of just the Chinese language that I feel like I noticed, I thought differently when I learned English. And there were some studies around this too, I think that there’s, I think in, I’m not sure exactly, I just have to go check up on this. But I think in Russian there are two different words for blue, there’s a greenish blue and a bright blue or something.
Second Hiring Secret: Growth Mindset
Lenny Rachitsky: I speak Russian but it’s like… I moved to the U.S. when I was 6 and so my Russian is not great. So, I’m trying to think of this as you say it, but keep going.
Peter Deng: Well, so then this is great. So, I need to get a way to validate this. But from what I remember, because there were these two different words for these different shades of blue Russian speakers who then learned English had an easier time distinguishing between these two shades of blue than, and a faster time doing so than people who had just grown up speaking English. So, I read some studies over there. And also there’s some other languages that don’t actually have a word for blue, I think. And then that’s actually really hard for them to distinguish over time. So, that really stuck with me and I think that it’s kind of rings true. So, when I, how I put it in practice, is that when I make slide decks, I gave a presentation to a class a couple of weeks ago and there were probably a total of 20 words on the entire slide deck.
And I spent hours obsessing over them because I really wanted to make sure I captured the right essence of what I was trying to say. And I think that crafting is really important when you’re working in product, because if you’re sitting down and you’re writing a vision doc or you’re writing a PRD, and if you don’t pay attention to the words you use, and you’re not intentional about it, those have downstream effects. People might misinterpret things, the connotations may not actually come through. And so I really am very careful about it because I think that there’s a multiplicative effect and a downstream effect for using the wrong word. And I really believe in that kind of language affecting thought thesis which is why I’ve just really, really paid attention to that.
How to Spot a Growth Mindset
Lenny Rachitsky: Mm-hmm. Yeah. And I feel like AI can help you with that too.
Peter Deng: Yes. Exactly.
Management Skills and Managing Up
Lenny Rachitsky: We had an episode-
Matching Is a Two-Way Street
Peter Deng: Well, actually, speaking of AI, actually that’s a really interesting point. I think it’s really interesting and kind of poetic and fitting that the breakthrough in artificial intelligence came from large language models. It’s interesting to me because there is, with every word in every sentence so much of the knowledge is encapsulated and shaped. And when ChatGPT does something really interesting, I tell people it’s oftentimes just writing Python code and interpreting it. And Python is a language yet again. So, I think that there’s something really interesting where like the condensation of human thought in language is related to the LLMs and the advancement scenario that we have today.
Lessons in Management
Lenny Rachitsky: I think it was Ilya on a Dwarkesh’s podcast where he was talking about, you may think LLMs are just like, “Oh, just predicting the next word, what’s the big deal?” But in order to do that, it has to understand the universe and everything in the world that has ever happened and existed and everything anyone’s ever written to predict the next word.
Internal Testing and Design Thinking
Peter Deng: Yeah, love it.
Operationalizing Empathy in Product
Lenny Rachitsky: Yeah. Okay. So, let me zoom out a little bit and shift a little bit to just product in general.
Why I Left Google for Facebook
Peter Deng: Sure.
How to Choose a Career Path
Lenny Rachitsky: You’ve worked at and built some of those iconic products in history. You worked at OpenAI, Facebook, Uber a Head of Product at Instagram. So, let me just ask you this question and see where this goes. What’s the most counterintuitive lesson you’ve learned about building products or leading teams that goes against common wisdom?
Peter Deng: I think one thing that, it’s a really hard lesson that I learned at Uber, which is sometimes your product actually doesn’t matter. And by product I mean the pixels you put on the screen or things that you build in your mobile app. And at Uber, I learned this because, it pains me to say this, but really the price and the ETA at Uber was the product. And I think a lot of times people at tech companies think of the product as just this digital manifestation, but looking at it from a holistic perspective, we humans consume the entirety of the product. And I think that was one of the things that I learned, the lessons that I learned that was really kind of hard hitting, that sometimes the pixels don’t matter as much as you think. And you fix a certain bug, it’s not to say that you shouldn’t fix the bug, but it doesn’t have as much of an impact as something that is more important to people like a price or ETA.
And this happens a lot in B2B products where it’s not just about how… It’s great that your product is well-loved by its end users, but does it make good business sense? Is one of those hard lessons I learned as a very bright-eyed, bushy tailed sort of design-based product manager going into Uber. I think the other insight that I had or other thought I had the other day was just the idea that so many of the tech companies today, this is kind of counterintuitive, so many of the tech companies that are most valuable today didn’t really start with any technological breakthrough. They were built on some kind of technological breakthrough and they ended up building a lot more technology. But really a lot of these companies, like Facebook for example, just put in the hard work, the elbow grease, especially in the early stages, to take essentially a database of human connections and build something valuable on top of it.
And that keep on polishing and iterating that product and coming up with new ones like newsfeed and photo tagging just kind of came out of just really paying attention to what people wanted. And some of the ideas are super simple and it’s not something that came out of the lab. So, Uber for example, took the fact that everyone had these GPS devices in their pockets, and they didn’t invent the GPS device, but they were able to take that and the fact that people had cars, and people wanted to get around, and there was a human need, and they just connected the dots, and put everything together.
And eventually, built a ton of tech to predict the right marketplace and pricing et cetera. But largely that’s a very valuable tech company. But it’s largely an operations company. And I want to give a huge shout-out to my colleagues there who run Uber Eats and Uber Rides from operations perspective. Because truly that was one of the biggest business model hacks that I’ve seen. And so I think that it’s Silicon Valley it gets lost a lot. It’s like, “Oh, this is a new tech company.” Oftentimes some of the most valuable ones are just the ones that are just building what people need on top of existing tech.
The Fail Corner
Lenny Rachitsky: There’s so much to say here. I love it. And this is coming from someone that led the Uber Rider product team and worked at Facebook and a Head of Product at Instagram. It means a lot coming from someone like you, not someone that’s not in product especially.
Peter Deng: Yeah, I mean, just to go further on the Instagram part, the idea was super simple. It was showing photos and visual sharing. But the craft that Mike and Kevin had in putting in the hard work to get the product just right, that’s what made it really take off. That’s a great example. I had forgotten about Instagram, but how could I? But it wasn’t anything that any other company couldn’t have done, but it was that product taste that Kevin and Mike had and conviction that there’s a certain sort of vibe, if you will, that people wanted, and building that and iterating, and look at it now, it’s a core part of our lives. Visual sharing, they really solved it.
Quick Lightning Round
Lenny Rachitsky: Yeah, I just had Mike Krieger on the podcast. So, it’s interesting, there’s two tensions here. One is just the product doesn’t matter in a lot of really successful companies. It’s secondary to the cars, the drivers, the GPS and the phone. And then on the other hand, there doesn’t need to be a technological breakthrough to build a huge business. It’s almost like if there’s no technological breakthrough, then the product matters. Facebook is an example. Basically, it’s like a database of connections, but what allowed an Instagram, what allowed them to breakthrough, and there’s classically competitors at the time. Was the experience, was it a lot better? And then maybe on the flip side, if the experience doesn’t matter, then it’s the breakthrough is on the operations and other… Does that resonate? Is that kind of what you’re saying?
Peter Deng: It does resonate. I think both have to be true. But also I would say that even if you did found a company that has a huge technological breakthrough. Very shortly, I think that the product experience will start mattering, because how long does that technological advantage last before humans wisen up to be like, “Well, this is not the product I want to use. I use it a little bit differently and this is more ergonomic for me?” Et cetera. So, I think that what you said is a beautiful summary. I also think that a point in time in a company’s history will also determine what is going to be more important.
My Life Motto
Lenny Rachitsky: This is, especially, interesting for companies building on top of LLMs and AI infrastructure, where you’re essentially saying, you don’t need to have some kind of technological breakthrough to build something valuable if you can create a really special, unique experience that unlocks the potential of this super intelligence.
From Product Leader to Investor
Peter Deng: I think that’s right. And I have some more thoughts on just sort of the companies that are building on top of LLMs that are just… That’s a slightly different thing I would say. I think that for them, having the right data, and that right data flywheel’s is so important.
How to Get in Touch
Lenny Rachitsky: Like proprietary data especially.
Peter Deng: Exactly. And the flywheel part is just, you can start with proprietary data, but the flywheel is really just sort of how do you continue to maintain that and generate that. And the second thing is, again, it’s the workflow. So, it’s the ergonomics of how does it actually integrate into people’s lives? And that is going to be more and more important.
Lenny Rachitsky: Well, let’s actually spend more time there because a lot of people are thinking about this. It feels like everybody’s trying to start a company these days with AI enabling so much more. And so I think a lot of people are just curious, where should they spend time? And so I think this is actually really interesting. So, what I’m hearing here is two things to think about to create any kind of moat, defensibility against, say, foundational models coming to your lunch or in other companies. What sort of data can you acquire that is proprietary and create a flywheel to generate more of that data? And then the other piece is how do you fit into a very specific, basically, vertical that you understand really well that fits into their existing workflow? Is that…? I’m probably right.
Peter Deng: Yeah. Well, it’s, again, this is something we can unpack for a long time. Because with any product that you want to build, there’s going to be incumbents that have distribution advantages. But I do have this thesis that there are certain-
… have distribution advantages, but I do have this thesis that there are certain products that will be able to break through those advantages of the distribution of the other companies, but you have to overcome a pretty high bar of your product has to be so much better. I think that’s one thing.
But yeah, I think that data flywheel thing is really interesting because the models will get really good at whatever data you show it, and that’s one of the things that people just think that AI is such a magic wand. But no, it’s like if it’s been trained on the right data, it’s going to do the thing that it’s been trained on. It’s very malleable.
Being very mindful of the data that you have access to to start your flywheel going and what you can do to keep on going with that flywheel is going to be a critical thing for anyone who’s starting a company today.
Lenny Rachitsky: Let’s make that even more specific. When you talk about this, I think about… The CEO of Windsurf was on the podcast and we talked a lot about how they’ve all this really unique data about which recommendations of code snippets people accept and reject and they actually launched their model I think based on that. Is that example? Any other examples to make this real?
Peter Deng: That’s a perfect example.
There’s some companies I’ve invested in that aren’t public yet that have their own take on that, which is really interesting to be able to take whatever activity is in their product to get smarter at the thing that they are doing, again, which is why I think the data flywheel and the workflow goes so hand in hand together, because if you are solving something actually valuable for businesses, for people, and there’s a lot of that attention that’s being paid to, a lot of work is being done through it, you’re going to have that edge.
This is where I see again startups in very different markets who have this insight, who understand this very deeply, and are not just trying to zero shot everything and be like, “No, no, no. This is how we’re going to build it to make the product genuinely useful so that it can get genuinely more useful over time.”
That is going to be amazing because as a consumer of any of these products, we’re going to benefit.
Lenny Rachitsky: What I’m hearing here is also if you don’t have proprietary data or unique data, you can still have a chance by building this flywheel where you collect that data through your usage.
For example, Windsurf, if they all built on Claude 3.5 and then now they have all this unique data and now they’re launching their models.
Peter Deng: That’s exactly right.
This goes back to something I might’ve mentioned briefly, but you got to have grit when you’re building anything. You got to be able to have that vision, have that clear direction, and be able to really go chase that. I think that’s really important.
Lenny Rachitsky: To make your example of distribution being overcomable, a great example I think a lot about, and we had CPO, turns out there’s many CPOs at Microsoft, I didn’t realize how many CPOs they had, and I asked her about, “Why didn’t Copilot…” The fastest growing companies in the world, Cursor, Windsurf, Lovable, Bolt, all these guys. Copilot was so ahead of these companies and these companies broke through.
While Microsoft has distribution, amazing talent, infrastructure, all the things, early first mover advantage and it’s to your point, they were just building products that were much better, Cursor, Windsurf, all these, Lovable, Bolt.
Peter Deng: I do believe there is a level of product craft that will make it so that it’s just worth it to switch or try something else. There are a few products out there that I see with this. I think Granola is one of them.
There’s so many distribution advantages that Google Meet has, that Google, Facebook started off, Microsoft Teams has, Zoom has, but they’re just these tiny little product craft delightful things that I really appreciate myself of like, “Yeah, they got it.”
They have these little edges, set it down just right, and they’ve really figured out a way to really make it so delightful that it’s like, “Yeah, I will install this piece of software. Yes, 100% I will talk to my friends about this because it is so life-changing.”
We’re starting to see that now. Again, before, I would say 18 months ago, it’s like, “Oh, well, who has the best model?” But then coming forward, it’s like really who has the best workflow and who has the best product, and we humans are just demanding. We want the best. And so when someone is going to come out and produce something that’s so well-crafted, I think people are going to pay attention.
Lenny Rachitsky: A couple of takeaways here is if you’re trying to build an AI startup, a few things you should be thinking about that gives you a better chance of breaking through and winning is what are your data flywheels where you collect proprietary unique data, how do you build something the craft comes through and people are wowed and want to tell their friends about it.
Granola is a great example. Clearly, Cursor, Lovable, Bolt, Rep, all these guys did that and then it feels like they just understand a vertical workflow really well and someone’s problem and solve that in a really unique way.
Peter Deng: Yeah. I couldn’t have put it better myself.
Lenny Rachitsky: Awesome.
Let me ask you, this came up in my chat with Mike at Anthropic and it’s along these lines. I was thinking about just what is product doing at Anthropic.
They’re building this basically a gigabrain super intelligence thing that’s going to know everything and maybe build its own experience in the future. And then there’s this product team building this layer on top to interact with this super intelligence gigabrain.
Peter Deng: What is the point? What is the value of that layer?
Lenny Rachitsky: You spoke to it a bit here of just there’s value in the experience and feeling native, but I guess let me just ask you that. Just where do you think product goes at a company like Anthropic, OpenAI where there’s just the super intelligence that the team is working on and there’s this UX on top?
Peter Deng: I think those companies have just such an advantage because you get to work in the same building as the researchers. I think that there’s that really symbiotic relationship, close partnership between post training and product where, again, more and more it’s going to be less about the raw intelligence, it’s going to be about the fine tuning of what the model can do that really resonates with people and what people want and also what the product trajectory is going to be. I think that you’re going to see that more and more.
I think this is less about Anthropic but more about OpenAI. I think OpenAI made a great move.
I am a huge Fidji fan. As soon as that news leaked that she was going to join, I texted her. I was like, “This is great. Amazing. Congratulations.”
I’m thrilled for her, for the company, for all of my friends still at OpenAI because it’s just going to be this amazing leader coming in.
I’m also thrilled as a consumer because some great products are going to come out.
I think really that close, tight-knit relationship at any of these large model companies between post training and product is going to produce some really incredible stuff.
Lenny Rachitsky: First of all, Mike actually said very similar things that the more-
Peter Deng: I promise you I did not watch that podcast.
Lenny Rachitsky: It hasn’t even come out yet so I believe you.
Yeah. He had this interesting finding where he put product people on UX product experience front-facing product and then he put PMs on the research teams and building models, helping models get better, helping researchers build things, and he found that all the leverage and wins came from the PMs working with the researchers, much less so on the product experience. And so he puts more and more PMs with that team.
Peter Deng: I’m so thrilled to hear that because that’s a little bit… It’s very validating because that’s what we did at OpenAI too. We were very closely tied to the post training team and it was because of that tight collaboration that you see some of the advances of ChatGPT getting better at so many things. It’s great. It’s awesome that we independently came to the same conclusion.
Lenny Rachitsky: Yes. It’s a good sign.
Okay, so we’re talking about startups, building new companies. I want to follow this thread a little bit.
Peter Deng: Sure.
Lenny Rachitsky: I feel like you’ve built more products from zero to one to scale than maybe most anyone else across all the companies that you’ve worked at. I’m going to do a quick rundown of some of the things you’ve done and I’m going to miss a bunch but let’s see.
You built and led the Facebook Newsfeed, the current version of it. You built the new groups experience chat and messages. You shipped the Messenger app as its own app. That was one of your projects.
You led UberPool low-cost rides. You launched ChatGPT Enterprise. You shipped voice and vision, memory, custom GPTs, just refreshing the whole design of ChatGPT. Many more things.
A lot of work at Airtable obviously. Also, Oculus.
These are just some examples in the intro. I’m going to try to go through all these things.
All that to say, I feel like you’ve seen a lot of what works and doesn’t work, building from idea from zero essentially to one to scale. So let me just ask you this question, what’s an important lesson you’ve learned about what it takes to succeed building something from idea to one to billions?
Peter Deng: Yeah. Thank you. That was a good trip down memory lane too when you read that off.
I think the first thing I would say, going from zero to one is different going from one to 100. When you are in the one to 100 phase, which is a lot of the time that I spent is the one to 100 phase, we quadruple Instagram usage in two years, that was very much a fun ride and there’s a bunch of other examples at other companies.
But when you go to one to 100, I think one of the things that you really got to take into account is that you have to plan your chess moves out in advance. You have to really think before you act and build systems that are going to let you go sustainably faster, because the zero to one is you’re trying to find that product market fit and then when you get to one to 100, you’re trying to make sure you can get to hyperscale as fast as you can.
I’ve been very fortunate to be along the ride of many of these products as they were going through that hyperscale. And the analogy I always like to use is that when you do that, you feel the G-forces. Some people are like, “Oh, yeah, I’m a pilot, I can fly at 35,000 feet.” But feeling the G-forces of takeoff of a rocket is very different.
One thing that I’ve learned there doing that a few times is you got to build the systems that help you move sustainably faster, and sometimes, you have to go slow to go fast.
Here’s an example.
In building the Newsfeed, the current version that we have today, it really hasn’t changed much from the time that we built it, I don’t even know, it was like 12 years ago or something, I don’t know the reason why it hasn’t changed much.
But I like to think that it’s because we put a lot of time and craft into thinking about the whole sharing loop and what are the key pieces of it and how is it architected, what’s the information architecture, and what does that whole flow look like, how does it go from posting something at the top of the page to showing up in the newsfeed to someone clicking like and then that notifications thing lighting up red and then that repeating over and over again.
I like to think that Newsfeed has stood the test of time, the current version of it, because we thought very carefully about how people wanted to interact and how people wanted to consume information and also, that whole loop. When that happens, then I think things are built to last. I think this is a case at a lot of different companies.
When I was at Uber, we had a bit of a spaghetti string code situation on the writer app, but taking a step back and re-architecting things of what are the core components and how do you actually make it so that the product selector can scale around the world.
Here’s a little known fact. Talk about grit and elbow grease.
Uber’s not just as simple as finding a ride. If you’ve ever been to another country, like in India, sometimes, there are no street signs, so you have to pick up in front of this mini mart or whatever it might be. There’s a whole team that worked on pickup and drop-offs. This was a large effort.
It sounds so boring but it was so critical to Uber being able to scale because pickup and drop-offs team thought about, “Well, how do you do it for venues?” That venues and finding that right abstraction means that you can have a scalable way to do pickups at airports and configure different venues.
Those systems when you take the time to build them in the one to 100 phase help you speed up massively and that’s how you get 4x users in two years.
Or on Messenger, we put a lot of thought into the infrastructure around push notifications, etc. We grew that product from zero to 4.7 billion messages sent per day in about two and a half years. I think it really requires that forethought in building the right systems.
Lenny Rachitsky: Let me follow that thread real quickly because that’s really interesting.
Essentially, what you’re saying is there’s a phase of once you find product market fit, and I want to actually ask you this before you start planning, when you’re starting to scale going from one to a hundred, your advice here is basically don’t move fast and break things. Don’t ship MVPs. This is the time to really think many chess moves ahead about what you’re going to need to get this to, say, a billion users.
Peter Deng: Yeah, yeah. It’s building the systems and then that systems thinking will carry you really far, or at least that’s been my experience and hopefully, you can find the same way but your biology may vary. But yeah, that’s exactly right.
Lenny Rachitsky: What’s your guidance on just when to do that? Because you build something, okay, well it’s working, there’s also this just like, “Okay, let’s just keep it going, let’s scale it as far as we can.” In your experience, is it… Just what’s the guidance on when to really step back and really think years and years ahead?
Peter Deng: Great question.
The first thing I’ll say is that it’s not a binary switch. It’s actually a ramp rate.
When I’ve led teams, I’ve always believed strongly in this portfolio approach. Famously, Google had the 70-20-10 portfolio approach. That may be the right thing for a more mature company, maybe it’s 50/50 if you’re a startup, but you have to think about this in a non-binary way and in a way, that’s about scaling up and when do you need to put more resources behind that.
Every startup is going to be different. Every product that you’re launching is going to be different. And then thinking about your portfolio approach and how much you allocate your time that would be my advice. It’s really dependent on the stage that you’re in.
I think that actually is a nice dovetail to my second thing, if I may, which is when you’re going from that stage of maybe one to five or one to 10, so not just fully one to 100, one thing I found to be very helpful is to measure everything.
This sounds, again, very simple but just like how you wouldn’t fly a plane without instruments, why would you run your products without understanding the instrumentation and how it’s doing.
One of the things I did in pretty much all the teams that I led, whether it was Instagram, Uber, Airtable, was all about… ChatGPT too.
One of the first things I did was always to build a growth team.
Building a growth team is really interesting because it actually is a simple razor, it’s a simple thing to think about. It’s like, “I’m going to build a growth team,” but then you’re going to uncover a lot of things.
You’re going to uncover how much stuff you have not yet logged and how non-rigorous you’ve been looking at your entire product.
It’s so funny because I’ve seen this movie so many times, the same movie so many times that every one of these companies where I remember walking into Instagram and I think asking Kevin and Mike, “So how many users do we have?” It’s like, “Well, we don’t really know.” And so it’s like, “Well, there are a lot and we don’t really know.”
When you build a growth team and you hire the right growth leader, I’ve had the pleasure of working with George Lee at Instagram, some of the early growth folks at Facebook, Andrew Chen at Uber, Airtable. I had the privilege of working with Lauryn, who is currently now leading growth at Notion. I’ve been very fortunate to work with some really amazing people on my team.
When you hire the right person, they start asking all the right questions because when the archetype of person who is a growth PM will be like, “Well, wait. Why is this happening? And let’s get the data on X, Y and Z thing.” That’s when you realize you don’t have X, Y, and Z thing logged and after you have X, Y, and Z thing logged, you look at the data, you’re like, “Wait. Well, why is that happening?” And then you’re forcing yourself to go deeper into the analysis of doing some analysis of like, “Well, what’s correlated with what and what are some hypotheses?”
Because growth leaders, growth product leaders are so into this experimentation side, it actually is this really easy thing to do is when you start building a growth team, it just begets all of the right questions being asked and then it starts turning into all the right behaviors of taking something you’ve been building, which seems like it’s working into a more rigorous system.
That’s zero, sorry, the one to 10 phase I would say that really sets you up for the 10 to 100.
Lenny Rachitsky: What I like about this growth team advice is that a lot of people think of a time to hire a growth team to we need to drive growth. What you’re saying is there’s a lot of second order benefits, which is they help you figure out what the hell’s going on and inform a lot of other things that are happening, people just actually understanding how things are going.
Peter Deng: Totally.
I think that the reason why growth team is the advice I would go with rather than to build an analytics team is because if you build an analytics team or a data science team, it’s possible that no one’s going to listen to them. It’s like, “Oh, I have these insights.” It’s like, “Well, no one really cares.”
But if you hire a growth leader, they are now tied to outcomes of driving growth, so they’re going to be the ones who are listening and asking more questions and really partnering with that data science team to make your entire product and business more rigorous. That just changes the DNA of your entire team.
Lenny Rachitsky: I want to talk about hiring, but is there anything else along these lines that you want to share of building new products, scaling products?
Peter Deng: I guess the last thing I would say is I want to make sure that sometimes in the pursuit of numbers, product folks lose sight of the importance of taste and craft. Maybe this is actually the dovetail into building teams, but you got to have the counterbalances.
It’s really important to give two people on your team different charges. One is like go grow the product and the other one is wait, maintain that design, that beautiful aesthetic, the craft that your product is known for. That tension is extremely healthy. I’ve seen this at Facebook. I’ve seen this in Instagram. I helped create this at Instagram, this healthy tension. Airtable, same thing, but just having… ChatGPT, same exact thing.
You have to have that push and pull on both sides to really stretch the gamut.
Lenny Rachitsky: That begs the question, how do you actually do that? You could talk about it, you could be like, “Okay, we need to make sure the experience is awesome but also grow this number. Here’s your goal.” How do you operationalize that? Is it a performance review? Attribute thing? Is it culture or something else?
Peter Deng: As a leader, you have to set up your team the right way. You have to really think about your team as a product and what are the various pieces you need to really stretch the gamut of what you’re thinking about.
The teams that I’ve helped build are… The most successful ones are a team of Avengers that are just very different, have very different superpowers, but together you as the leader are the one who’s helping adjudicate any differences or any disagreements but you know you’re getting the best outcome when everyone’s pulling and obsessing over a different thing. And that’s important.
It’s important to create your balance and really increase the space that you’re looking at and create those healthy debates.
I think a lot of people overlook that. I think some people think of people on a team as warm bodies to do a job, but my philosophy has always been to think about, “Well, what does the company need to be successful and who’s the best person who spikes at that one thing and how do I make sure that we get that person and how do we make sure we get the other person and the other person?”
It’s almost like you’re playing an RPG where everyone has different sliders and you have to create this super team where everyone actually spikes in different ways.
That is something that I’ve had a lot of success with in terms of when you create that environment and you create that vibe, you’re going to get a lot of mileage out of that team.
Lenny Rachitsky: That is a really interesting answer. It’s not one I’ve heard before. Essentially, it’s not create the right incentives, it’s hire people that naturally see the world in a certain way and that creates a balance and a healthy tension between say a PM and a designer and an engineer.
That is really interesting because that feels a lot more sustainable than like, “Here’s your goal. But also when your goal is make sure the experience is great and people support tickets are down.” It’s just like naturally, they need to want this to happen.
Peter Deng: Totally.
Actually, I have a framework around… I think there are five different types of product managers that has held true.
This is a framework that just came out of a random jam at Uber when I was talking to some of my colleagues there. We formulated this in terms of helping with hiring practices.
Everywhere I’ve gone, I’ve also been best friends with the recruiters because honestly my whole thing is got to build the right team. So we have to really partner very deeply.
At Uber, we developed these five archetypes of a PM. To this day, I still think it’s actually exactly true and it still holds true to this day, but is that interesting? You want me to go into that?
Lenny Rachitsky: Absolutely. I’m so excited to hear what these are.
Peter Deng: These are the five that I’ve found to be most enduring and actually the most different.
When you talk about… I love the way you put this, Lenny, which is when you hire the right people and they’re naturally motivated by different things. These are the five that we came up.
Number one is the consumer PM. This is the person that’s half designer, half product person, really obsessed over the details. “Is it delightful? Is it crafted enough? Oh my goodness, this is three pixels off. I can’t stand it. This is driving me nuts. Why is this so complex?” These are the people that you think of as sometimes the criticism PM is the consumer PM, but that’s just one type.
Another type, just on the other side we’ve talked about before, is the growth PM. These people are half data scientist, half product person, they are wired to think numbers first and they have this air about them that’s like the best ones do, which is like, “I’m really skeptical. Show me the data. Let’s run a test and prove it. I don’t believe you.” I start with these two in the framework because they’re actually really different. One, it’s like, “I have vibe, I feel the vibe, this is better,” and the other one’s like, “No. I don’t believe you. We should test this and prove it.” That’s a really healthy tension.
I love having two people in a room debating that. I’m like, “Great. We are going to get some good things done and we’re going to move the product forward.”
The third type is what I call the GM PM or the business PM. These are half MBA, half product person. These are folks that are naturally wired to start with the business model and think about, “What are the margins? What are the opportunities? Where’s the value being created?”
We had a lot of these at Uber and they were the marketplace PMs and they’re just like…
I loved working with them because their minds just worked differently. They just thought about problems from like, “Well, what is the incentive here?” This is a fascinating type of mind to work with.
Another one I found, it’s actually more nuanced than you think, is there’s a certain archetype that I call the platform PM, which is someone who’s really deeply wired to build tools for other people.
At Uber, we had internal platforms for messaging or for building internal tools.
Oftentimes, these folks are overlooked but it’s actually a really deep wiring, because these are the people that are going to build the systems that are going to make you go faster. And that’s what they love doing.
The last one, I would say, I used to call it an algorithms PM, but now in the world of AI, I’m going to rename this to research PM. These are half researcher, half engineer, half product person. These minds are amazing.
Basically, they think traditional Google search algorithm PM but nowadays, it’s like who are the people who really have that product taste but deeply understand the tech and the way the models are trained to go and affect that and build the most amazing product.
Those are the five.
I still think to this day these hold true, and we might have been onto something the day that we brainstormed this at Uber but, yeah, I’m curious to hear your feedback.
Lenny Rachitsky: This is great. As you’re talking, I’m just like, “Here’s that person, here’s that person. Okay, they fit here.” This super resonates.
Ever wonder why customers drop off before converting or why some pages perform better than others?
Contentsquare takes the guesswork out of digital experiences, giving you real-time insights into how users interact with your site or app. With AI-powered analytics, automatic frustration detection, and clear visualizations, you’ll know exactly what’s working and what’s holding your customers back. Whether you’re optimizing an e-commerce checkout, refining a B2B lead flow, or improving a mobile app experience, Contentsquare pinpoints exactly what needs fixing and why.
Contentsquare powers better customer journeys across 1.3 million websites and apps. Discover the insights you’ve been missing at contentsquare.com/lenny.
Just to summarize, there’s consumer PMs, growth PMs, business/GM PMs, platform PMs and research PMs.
Peter Deng: Yes.
Lenny Rachitsky: A lot of people call them AI PMs now. I feel like that’s the term that’s really [inaudible 00:55:51] now.
Peter Deng: You have to evolve with the times. Yeah.
But also the other part of the framework I find interesting is that everyone has a primary one and a secondary one.
It’s like one of those personality tests. Maybe we did this just because it was hard to pigeonhole people and I myself don’t think I was pigeonholable, but I do think that people lead with one type of thinking and then also have the secondary thing that keeps them in balance.
If you believe that and you apply it to your team, I’m curious to hear from your listeners if this does resonate or not. Maybe this framework will help you realize that you’re missing someone that you should be not missing.
Lenny Rachitsky: What was your archetype when you were a PM?
Peter Deng: That’s the other thing with personality types is the ones you hear. You’re like, “This is me. I own this.”
There’s no doubt about it. I am a consumer PM and also a growth PM. That’s my primarily consumer… I can’t…
This is what I told you about the other products I’ve loved. I can see the details that people put into it and I so appreciate that. But at the end of the day, it’s like, “We got to measure things.” That’s what I am. But again, everyone’s different.
Lenny Rachitsky: I love your point about how a lot of people think of PM. They hear that first example and like, “Oh, I guess that’s what I need to be, because that’s what everyone talks about when they’re amazing product managers.” But you’re saying there’s many other ways to be a successful PM.
We did a personality test at Airbnb when I was there, and one of the biggest takeaways was it’s like this color test and you get a color green or yellow, red, and the team was all over the spectrum. It was a really good reminder just you can be a different type of person and still be really successful in this role of PM.
It’s probably because of these different archetypes and different needs and roles of PMs. There’s this word product manager but there’s many things that PMs do.
Peter Deng: Also, as an investor now, it’s really important to see the fit of the founder to the market because if you put a consumer PM into a really boring regulated industry, they’re probably going to get frustrated and they’re probably not going to see it through. Whereas there’s people that you look at the pitch and you’re like, “Wow. You are really passionate about this-”
… pitch and you’re like, “Wow, you are really passionate about this problem, and you really care about building tools for others, and this is exactly,” this is the Twilio PM or whatever it might be. “You’re a perfect fit for this business and that’s awesome,” right? So I think, yeah, I love what you just said in the summary, because I think there’s no one way to be a PM, and I think this is, hopefully this framework will give people a little bit more space to express who they really are.
Lenny Rachitsky: I’m curious if other functions also have these sort of archetypes, like designers and engineers, but we don’t need to get into that. How about if you’re listening to this on YouTube, leave a comment of which of these archetypes you think you might be. What’s your primary and secondary? I’ll read them again. Consumer PM, growth PM, business/GM PM, platform PM, research/AI PM?
Peter Deng: Love it.
Lenny Rachitsky: Okay. I want to talk about hiring. So this actually came up a lot when I was chatting with folks that you’ve worked with, especially Nick Turley, who’s head of product at ChatGPT, who we’re trying to get on the podcast. Because-
Peter Deng: Yes.
Lenny Rachitsky: … that’s an-
Peter Deng: He’s awesome.
Lenny Rachitsky: That’s what I’ve heard. So he told me that the current head of engineering, the lead product engineer, the head of design and head of marketing at ChatGPT are people that you hired. Also, many of the people you hired have gone on to do incredible things. You’ve shared a few of those names, many of them have been on the podcast, which is the ultimate measure of success. So let me just ask you this, what’s one thing you look for in people you hire that you think people sleep on, that you think people aren’t paying enough attention to, that helps you find amazing stars?
Peter Deng: That’s really flattering to hear that from Nick. Nick is one of the best people I’ve worked with, period. In fact, I want to just do a quick shout out. Folks at OpenAI are pretty much the best people I’ve ever worked with in my career. When I took the job, I told the team, “This is going to be my last operating role, and I’m going to leave it all on the field, and I’m just going to go all out.: And basically I spent probably as much time, if not more time on recruiting and building the team as I did thinking about the product. And this is going back to what I said earlier about, I think you got to bring the right people together to have a huge impact. And oftentimes leaders overlook this and they’re like, “Ah, it’s just a warm body,” but truly, people who have strengths in certain areas compliment others with strengths in other areas. And when you build that team, amazing things happen. It’s the best investment you can make. It’s going to pay off so many dividends.
So I think that’s my opening salvo in terms of you got to get … Everyone who’s listening out there, you got to make sure you look at everyone in your team, you look at what you need, and you have to get the best in each. And truly, in my farewell dinner at OpenAI, I think I closed with just, “Look, I don’t even know what I would do after this, because all the best people I’ve worked with are here.” We have Ian Silber running design there, Thomas Dimson, Joey Flynn, Ryan O’Rourke. Nick Turley was an amazing I met there. Joanne, I mean I have so many people I’m missing, but Coley on product marketing, Antonow on the marketing comms side, [inaudible 01:01:07], the list goes on. Product operations is stellar. I’m so proud of, honestly, the team that I built there more than the products. So I just wanted to say that it’s a big thing that I really care about, and I hope more leaders think about that too, is really be mindful of putting your team together, and thinking about that as a product. And you have to really craft that. You have to really care about the team. So-
Lenny Rachitsky: Just to double down on that point, actually, before you get to the next tip here, I just love this answer, which is, if I were to ask someone, “What’s your hiring vice? What do you look for that people may not be looking for enough?” Most of it would be like in that person, here’s what you need to focus on, and here’s the interview question. But kind of your broad answer so far is it’s not actually about the person, so much as what is the team going to look like, and where do we need spikes? Where do we need to balance out the composition of this Avengers that we’re building?
Peter Deng: Totally, totally. That’s exactly right. And so that being said, I guess I have, I guess, on brand, I have two things I want to share about hiring the right team. I have this saying, I actually have this doc that I’ve taken around various companies called the PXD API, which is like, “Here’s how to work with me.” And in it, there’s a saying that I have, which is what I really optimize for for everyone that I support and everyone I hire, which is in six months, if I’m telling you what to do, I’ve hired the wrong person. And it’s just kind of served me really well on three different levels. Number one, it’s a reminder for myself when I’m either hiring, or looking for the person, is to keep my bar super high and just not settle. Because if I do, most likely in six months, it would not be true that I would be able to let this person run, and I would still be telling them what to do, which is not what I want. That is not my desire.
The second sort of effect of that is that it’s … I say that to people when they come on the team or as we’re making the hire, because it communicates to them that that’s my bar, and that’s how they know they’ll be successful, and something to kind of work towards.
And the third thing is kind of a joint thing for both of us, which is it kind of gives us, it helps me and the person operate on a different level, where the goal is not did you hit this OKR, did you hit this goal? The meta goal becomes, hey, are we calibrating enough? Are we actually getting to a spot where in six months, you’re the one telling me what needs to be done? Are we getting there, right?
Because then, if that’s the framing, every mistake that is made or whatever on either of our parts becomes a learning opportunity in terms of like, well, how do we grow from this to where we want to be in six months? And how is it possible that I, as a manager, can do the right things to set this person up for success, so that I don’t have to be involved in six months?
And I think that those three things, and being able to have that second-order effect of this simple razor, in six months, if I’m telling you what to do, I’ve hired the wrong person, it puts pressure on me, it puts pressure on the person, and it creates this really interesting environment and this kind of safe space to really think about, are we heading towards that goal? And again, every place I’ve been at, as much as I’ve loved building the product, I’ve taken so much pride in building the team, and it’s just been so much of a pleasure. And I think this is one of the two secrets that I have here.
Lenny Rachitsky: This is so good. I have a follow-up question, but just to point out why I think this is so genius is that there’s kind a assumption here of this person, you can trust them. So there’s like, do I trust this person? Do I feel like they’re going to be proactive? Do I feel like they’re going to have correct insights, essentially taste and gut feeling? It’s like the layer below this question, which is great. And also just this autonomy, it feels like autonomy almost implies so many important traits of somebody that you want to hire. And I love just how simple this question is for both you and them to [inaudible 01:05:35]-
Peter Deng: Thank you. And really with that autonomy, I love what you said about autonomy. Because truly, as a leader, as a manager, your goal is to scale. And if this simple statement is not true, how are you able to build the best company, the best product?
Lenny Rachitsky: So here’s the follow-up question. Is this mostly for leaders, like say a head of product at ChatGPT, say, someone’s not a CPO, they’re just like, I don’t know, a manager of a PM team, is there a version of this that you think might be useful to them, or is this mostly for leaders?
Peter Deng: I think this is for everyone. I think it’s for everyone who is a manager. Because if you’re going to be a successful manager at any company, or a leader at any company, and if you’re starting as a line manager, or whatnot, and you’re kind of wanting to grow, or even just wanting to … If you’re early at a company, you have so much institutional knowledge. And so getting more leverage in terms of being able to pass on the wisdom that you’ve learned is so crucial into being successful that I think every manager should approach their reports with this. Because truly, it’s just good for everyone. It’s good for the company to have more kind of leverage and scale. It’s good for the person who is being brought onto the team, because they know what success looks like, and it gives them a path to keep on growing. And it’s great for you as a leader, as a manager, to be able to basically scale up the entire expertise of your team.
Lenny Rachitsky: I imagine you don’t even need to plan to not tell them what to do. It’s just a good lens into, are they going to be amazing? Even if you plan to be telling them sort of what to do.
Peter Deng: Yeah, exactly. The other thing is, again, in your interview process, you kind of end up looking for these insights, and you look for the behaviors of like, oh, are they actually going to be potentially able to achieve this in six months? And that’s going to give you a really good lens on the picking side, not just the development side as well.
Lenny Rachitsky: Peter, what’s your second secret? This is one-for-one.
Peter Deng: Yeah. Okay. The second one I’d say is, I feel really strongly about this, which is the area that I look for most is growth mindset. And I actually came to this some point in my management career at Facebook, where I did make a mistake and hired someone who just didn’t quite have that growth mindset. And it was really difficult, because the way I say it’s like, “Look, I don’t have time to sugarcoat any feedback,” and frankly, the best people I’ve worked with are the people who come into one-on-ones with me and yell at me and tell me I’m messing up. I love that, because there’s nothing left unsaid, and we’re able to kind of move the ball forward of, “Hey, how do we get better from this?” And I feel like growth mindset’s one of those things, Lenny, that it feels really hard to teach at a certain age. And this is really important to me and my family, I expect growth mindset of myself, of my kids, my colleagues at work.
Because I think it just creates this environment where everyone is open to what’s the one thing I can get better at? And that whole get 1% better every day can become true. And it’s funny, whenever I go to teams like ChatGPT or Uber, when I’m always the final interview for someone in my org, and I partner with recruiting on developing the rubric, I always insist on doing the last interview. And I do … not product sense, I don’t do design, I don’t do execution, I don’t do metrics. I only do growth mindset.
And it’s kind of like, well that’s crazy. What about all of these other attributes? I’m like, “Well, I’m pretty sure I can trust the other people to assess the other attributes.” But I think the growth mindset thing is so important to me, that we build a org where people are self-reflective, and want to get better, and take that feedback, and give that feedback. And it just is this meta unlock that I found to be true. And really, if you don’t have growth mindset, and you’re not open to feedback, and you’re not open to learning, then that’s the meta blocker. At that point, it’s hard to get feedback, it’s hard to onboard to a new skill. It’s hard to develop in any sort of meaningful way. And so I found that to be the really critical piece.
Lenny Rachitsky: That’s a big deal what you just said there, that essentially as the CPO, head of product, big product leader at a company, your interview is not like, “Are you an amazing product manager? Do you have products taste,” things like that. It’s a growth mindset.
Peter Deng: And I just want to clarify, it’s because all the other things have been interviewed by the designer, by the engineering lead, et cetera. And that’s where the previous principle comes into play as well, in terms of, I do trust my team to go and assess those people, but the one thing that I care so much about is growth mindset. And that’s kind of the thing. And to be honest, I do do a little bit of a sweep. So if we got weak signal on one of those areas, I’ll do it. But the pure focus of my last interview is going to be on growth mindset.
Lenny Rachitsky: Okay, well I need to ask you what that looks like. But before I do, when you talk about growth mindset, I have this image of Mark Benioff on the podcast, and I asked him, just like there’s so much changing all the time. It’s such a crazy world to be leading a company in this world, where just, everyone’s disrupting each other, AI’s changing everything. It’s just moving so fast, every day there’s a new breakthrough, and you have to keep track, and just like, how do you deal with that? And he’s like, “You should be thinking, ‘Good. This is amazing. This is the best time to be building. There’s so much opportunity, so exciting. This is what we want.’”
Peter Deng: Exactly.
Lenny Rachitsky: “Good.” I just remember saying like, “Good.”
Peter Deng: I love that.
Lenny Rachitsky: And I feel like that’s the epitome of growth mindset.
Peter Deng: Yep, absolutely.
Lenny Rachitsky: Okay, so let me ask you just how do you tease out a strong growth mindset? What are some ways?
Peter Deng: Well, good thing I’m not an operator anymore, because I’m going to give away my interview questions, so no one can cheat on this. I feel like this is another reason why this is such a great time to do this podcast. The question I asked has been the same one I’ve asked for years. And you can really kind suss it out from this, which is I asked them, think about one of the biggest mistakes you’ve made, truly, the more painful the better. And tell me what the mistake was. Describe to me the situation, and tell me actually how you actually think differently now, work differently as a result. How has that turned into a core principle of yours, et cetera.
And I give them a moment to think about it. Sometimes I even share some of my mistakes, if need be. And it’s interesting, because I’ve asked this question so many times, I can smell the BS if they’re not being authentic.
It’s kind of like, “oh, I’ve worked too hard,” or, “I did this thing.” And they’re really not being that … You can tell the vulnerability that people are willing to express. And I reciprocate with that, if they ask me what mine is, I will tell them what it is. And then that’s the vibe.
But what ends up happening is there’s multiple reasons why this is really interesting. One, you get to get a sense of how reflective they are. And there’s one woman, I was chatting with them, we actually went on for an hour, because she was just educating me on this amazing problem that she had made this mistake on, and how it changed the way that she worked, and the company worked. It was just incredible. And you can sense the passion, you can sense what’s genuine. And then there are always, once in a while those things that people are just very, a little bit more defensive and not willing to open up. And it’s safe. It’s a one-on-one setting, so it’s a safe space. And it’s also, I don’t think it actually selects for or against introverts or extroverts. I think at that point it’s really genuine. And the second sort of order effect there is, if they end up coming on the team, you’ve already had that moment. You’ve already had that moment where you’ve just already said, “Hey, this is where I really messed up.” And guess what? It’s all okay. It’s not a loss, it’s a lesson. And so it just sets a different tone for your working relationship. So again, I’ve never A-B tested this, so I can’t tell you if this actually, works or not, but I found it to be very helpful in the style that I work in, to be able to have that level of connection, whether it’s with a direct report or somebody in New York.
Lenny Rachitsky: What I love about this answer is it’s very much like Fail Corner, which is a recurring segment on this podcast, and I might tweak Fail Corner to be even closer to this question. Okay, so let me summarize these essentially two questions that you’ve found to be really helpful in finding these superstars that you’ve hired over the years. One is you ask people in six months, “If I’m telling you what to do, I’ve hired the wrong person.” Or I guess, how do you say it when you say it to someone? Just like, “You’re probably the wrong person for this?”
Peter Deng: Well, it’s actually framed a little bit differently. So there’s five different part of my API, or just how to work best with me. There’s five attributes of people that are most successful who work with me and I love working with. And one of them is framed as that you’re telling me what to do, not the other way around.
Lenny Rachitsky: Six months after joining.
Peter Deng: Right, right. And then I follow up with, “In six months, if I’m still telling you to do, I’ve hired the wrong person.”
Lenny Rachitsky: Got it.
Peter Deng: I think, that’s how I frame it.
Lenny Rachitsky: Okay. By the way, you should open source this PXD API doc.
Peter Deng: I would love to. I think now I got nothing to hide. I’m just like, “Here, I’m an open book.” So maybe we’ll do that at some point. You’ll make me brave enough to do that, maybe after this podcast.
Lenny Rachitsky: So you may find a link in the show notes for this podcast to the doc.
Peter Deng: If I’m brave enough.
Lenny Rachitsky: Okay. And then the other question you ask is, “Tell me essentially a story of when you failed, a product that you launched failed, and how that changed how you behave, how you think about product, how you operate.”
Peter Deng: Yeah.
Lenny Rachitsky: Amazing. Okay, great. Okay, let’s talk about management.
Peter Deng: Sure.
Lenny Rachitsky: So this came up, so I talked to a bunch of people that have worked with you, and interestingly, one of the most recurring themes, it wasn’t about AI, or … Hiring came up a bit, but it was actually mostly about how skilled you are as a manager. And this all has already come through in a lot of the things we’ve talked about. So I want to talk about a couple things here.
Peter Deng: Sure.
Lenny Rachitsky: One is someone that you worked with at OpenAI, Joanna Jang? Or is it Yang-
Peter Deng: Joanne? Joanne.
Lenny Rachitsky: Joanne. Joanne Jang, or Yang?
Peter Deng: Yeah, Jang.
Lenny Rachitsky: Jang. Okay, cool. You worked with her at OpenAI, and she shared a couple things that I think are really interesting. One is that you had a profound impact on her career by teaching her how to manage up more effectively. And you did that by teaching her a really simple phrase that she just says and uses. First of all, do you remember what that phrase is?
Peter Deng: I’ve said a lot of stuff, and I’ve kind of forgotten. I tend to forget what I say, so you might have to remind me.
Lenny Rachitsky: Okay, so she said “Say you’ll do the thing, do the thing, say you did the thing,” as a skill of managing up. So just talk about that, just the power of that and what that’s all about.
Peter Deng: I mean, look, I learned this from my time at Uber, from Jill who runs PR, comms, and policy there, and she used to have this saying, which is like, “Repetition doesn’t spoil the prayer.” It’s just a natural thing where people are busy. So whether you think about managing up or even managing the entire org, if you don’t repeat what your goals are, if you don’t repeat what your vision is, if you don’t repeat the thing that you feel strongly about what you’re doing, whether it’s maybe to your manager, one, I think you might lose sight of the thing that’s important. And I think this is where it’s a little bit about behavior. This is another language affecting thought thing. By giving this phrase to Joanne, maybe it was just like, “Hey, let’s just be very intentional about what we build.” That becomes a constant reminder.
And it also has this other effect, where if you’re saying, “This is what I’m doing,” and then that’s a thing that your manager’s like, “Wait, we don’t need to do that anymore,” you can have a conversation about that. As opposed to just doing the thing and not saying that you’re doing it.
So let me take a step back. So one, say what you’re going to do. And then in that exercise you’re going to be able to calibrate with your manager, again, with anyone, what is it that we’re going to do? And I think the words are really important here, going back to what I said earlier, so figuring out what is that goal, and crafting that to really pack the most punch and the densest of concepts. And then you’re telling them that you’re doing it, which that’s the second phase, which is like, in your one-on-ones or in your team all hands, you’re saying, “This is what we’re doing.”It’s a great time to reaffirm you’re doing or invite the conversation that this is no longer the thing to do.
And you got to tell them you did it. So just close the loop, just be like, “Okay, great, this is now done.” And I think that’s, again, it’s one of those really pithy phrases that has so many second-order effects that are behavioral, almost. And this is a little bit of a hack in terms of helping people. It’s funny that Joanne thought of it as managing up, which it is, but in my mind it’s almost like this is how we operate, and this is how we’re successful to stay on task, stay on goal, and be able to revisit the goals that we’ve set when they no longer are relevant.
Lenny Rachitsky: So the phrase again is say you’ll do the thing, do the thing, and then say that you did the thing.
Peter Deng: Sorry, one more time. The way I would say it is, say you’re going to do the thing, say that you’re doing the thing, and then say that you did it.
Lenny Rachitsky: This also works for presentation advice. So this came up, I don’t if it was Guy Kawasaki or someone, had a very similar phrase that was for how to present well, which is tell them what you’re going to tell them, tell them, and then tell them what you just told them.
Peter Deng: It’s possible that I might’ve incepted it from there. So I take no ownership over this phrase. I will just say that yes, I did repeat it.
Lenny Rachitsky: This is great. And I love that this isn’t just managing up advice, it’s just operating advice for everyone. And there’s an implication of, the last part is just make sure people know what you did, almost make sure that you get some credit, and people understand the impact you’ve had.
Peter Deng: Which is important. I think there’s a lot of people who are kind of introverted, and don’t want to draw attention, and don’t have the hero complex. And I think that those people tend to get lost in organizations. So if that describes you, just remember to say what you did.
Lenny Rachitsky: There’s another management trait that Joanne shared that I want to spend a little time on, which is you’re very good at helping people understand that they can lean into their strengths, and not feel like they need to fit into a certain box. She shared that you basically helped her create almost a new role within OpenAI that wasn’t even a thing before. So just maybe share that example, and then just talk about why this is important, how you think about this.
Peter Deng: Well, I love that we’re talking about things that Joanne are telling you, because Joanne’s really special. I got to just take a moment to give her a giant shout out. She is the only person that I’ve worked with that has as much technical depth as she does have product taste. And I just want to pause there. It’s just truly special. I feel entirely privileged to have the chance to cross paths with her at OpenAI. I learned so much from her. Again, talk about not telling you what to do after six months. She was telling me what to do from day two, and I loved it, because she was so technical, and she has this taste and those two things are very rare to find together. And with Joanne, because she was so special in that way, and I spotted that, I was like, “Wow, I’ve worked with so many PMs and just like, this is very unique.”
It felt like we had to find a way to craft this. And sure enough, I was like, “Hey, can you just write up a job description of what is this thing? Because there’s something magical here, but I don’t fully understand it.” I don’t think any other person really thinks of things this way, and think this might be a big superpower for OpenAI. Let’s codify it.” And again, going back to my language being a really important thing, I think the exercise sometimes of writing things down, of things that you intuitively feel, give you an artifact that can kind of communicate with somebody else. So in this case, Joanne writing down some of the things that she got really excited about, helped me really understand that. And I was luckily in a position where I can basically say, “Look, let’s create this role. Let’s create this role and have you lead it. And I think this is going to be great for the product if we’re able to codify it.”
So I don’t think I did anything special. I was just following my instincts, and just following her lead. Again, I’ll be clear, I did not author that document. My recollection, she did that. So she did all the hard work in all of this thing, and I don’t want to take any credit for it. The only thing I did was just gave her a little nudge of, ” I think there’s something here. Can you just take a moment to go and write this down?” And when she did, it was just like, “Okay, this has got to be a role and you have to be the leader for this function.”
Lenny Rachitsky: What is the actual role she ended up in? I think that’d be really interesting to share.
Peter Deng: The role was model designer, and it was just a really interesting way that she framed it. And I know this role probably exists in some incarnation in other foundational model companies, but the way that she described it, and the things that she found to be the spikes required, led us to hire our first two model designers after running a search. And they were just perfect fits for the team. And that, I think, is largely a big secret as to why, at least, I’m biased. I love ChatGPT so much, and the way the model comes off, and the vibe of the model, is largely because of this technical plus taste role that she has created and she’s leading.
Lenny Rachitsky: I love one of the interesting takeaways from this is as a leader is just pay attention to what people are really, really excited about, and then take the step of, let them try to describe it very clearly in a doc. Coming back to your point about the power of language and words is just like, “Okay, tell me exactly what you’re thinking and let’s jam on it, because maybe there’s something here.”
Peter Deng: Yeah.
Lenny Rachitsky: Is there anything broader here about just leaning into strengths that you find just … There’s a lot of people, there’s all this debate of should I just work on the things I’m terrible at and that’ll make me better, or should I find the things I’m amazing at and just get better at those things? Any thoughts there?
Peter Deng: I genuinely believe that fit is a two-way street. And so what you are passionate about, what your strengths are, you got to really find the right company, the right role for you. And I think there’s a lot of force fitting that people want to do is to fit into a certain archetype. I’m glad we talked about the PM archetypes. Hopefully that frees people up to really lean into what they love. Because life’s pretty short. It’d be great if everyone would find the thing that they really wanted to do, and be able to lean in and do that. And I think the optimist to me is also why I’m so excited about the time and age that we’re in right now, because there’s so many different companies popping up. So there’s something that really resonates with people.
I mean, take a look at just what we’re doing here, it’s like, podcasting was not a thing 20 years ago. It was not a thing. But now, we are able to have these amazing tools and platforms that allow people to really express themselves, and really, what really truly brings them joy and makes them happy, and also brings a ton of value to the world. So I think that, yeah, I definitely believe in leaning in strengths, and I think that as hard as it may be, sometimes you got to look at where you are right now, and is this the thing that you really want to do? Or is there something else that’s drawing your attention and drawing you towards that?
Lenny Rachitsky: There’s another management oriented question I want to ask you. This came from Eric Antonell, who apparently has worked with you for 17 years across a bunch of different-
Peter Deng: Yeah, off and on for 17 years. One of my biggest mentors and friends, he’s amazing.
Lenny Rachitsky: Okay. So he’s like, “You need to ask this question.” So the way he put it is you’ve hired, managed, mentored many, many, many product people, some junior, some senior, across so many different cultures, and he’s just like, “We need to learn something from your experience doing that,” in terms of what you’ve learned about what it takes to be a really successful product person, whether it’s being successful in building product or career-wise, what’s just a nugget that you learned from seeing so many different types of people, and cultures, and seniority.
Peter Deng: I think for a product person specifically, it’s really important to obsess over the details of craft. Because ultimately, you’re crafting a product. It’s important to obsess about the details of craft, while simultaneously having the perspective and wisdom of which details don’t actually matter. I’m going to pause there and just kind of try to-
I’m going to pause there and just try to unpack this a little bit because at the core of being a product person, you’re like, oh, I want to build something that people love and that’s the job and that’s what draws people to be product people is that you have this desire to build. And I think that I’ve been involved in enough teams where I, myself, and when I was really young and coming up as a product person, I would just get obsessed over these little details and I realized afterwards that we’ve just wasted a bunch of time on something that didn’t actually matter. So I think that dichotomy is somewhat interesting and beautiful to me because it capsulates both the core of what the ethos of a successful product person is, which is you really have to care and you have to give a crap about the product that you’re building, but you also have to have the perspective and business know-how to understand where do you apply your time and where do you apply the care there?
And I myself feel like I’ve gone through cycles. Everything that I’ve done, I’ve gone super deep and really obsessed and then I take a step back and I’m like, wait, actually I was missing something and this other thing was more important, right? I’ll give you an example. I’ll use the Uber example here as what I said that the digital product didn’t really matter and it’s all about the price, the ETA, one of the products that I’ve built at Uber, which is Uber Reserve, right? It’s the simplest of things. Going back to what I said before, sometimes the best products is the simplest of things. But the problem that we were trying to solve is that everyone has this. You have a 6 AM flight, and are you really going to wake up at 4 AM and request an Uber and hope that there’s enough Ubers and the person’s going to come?
Because if you do that, you’re not going to sleep well and you’re going to wake up every two hours and you’re probably going to miss your flight anyway because you’re going to fall asleep or whatever. And so there was this insight of like, okay, there’s a whole mismatch between what people really want, which is the peace of mind that their car is going to be there and guess what? I’m willing to pay for that. And so we built Uber Reserve, which it was the simplest thing, which is like, oh, just go ahead and say what time your flight is and we’ll work backwards or even just tell us when you want to get picked up and everything about that product we crafted what really mattered for the user, which was the peace of mind. So if you go there and you say what time your flight is and your pick-up time or whatever, I think that the product is… It hasn’t changed that much since I was there.
It would tell you, oh, this is cutting it really close. You may not make your flight. It’s like, wow. Again, that was put in there because of the principle of peace of mind. And on the other side it’s like, well, what do drivers need? They need to know you’re not going to cancel and all this other stuff. So you’ve got to think about the driver incentives too. So it was a simple idea, really proud of the team for figuring out all the intricate details, did some testing, and last I heard from folks internally, this is a $5 billion a year business now and one of the highest margin ones, and I’m really proud of this because it came from the idea of let’s focus on what actually matters, which is that peace of mind and how many people really need it in that moment. So I think that’s the best story I can tell.
Lenny Rachitsky: That’s an awesome story. It connects so many of the things you’ve talked about. One is just it may not be the product that really matters, and micro-optimizing the experience is not going to move the needle when there’s something else that’s more operationally oriented, but there’s always going to be a product component if you’re building it for freezers. The other piece that I think is interesting here is… Well, there’s two. One is just it connects back to your point about the importance of autonomy of product people is just I feel like you’re like, here’s the team, here’s what I’m told to work on. And then you’re like, oh, but this thing is actually the problem we need to solve and let’s just build a new product around it. And then there’s a whole story I imagine of you getting buy-in and all that stuff.
The other thing this connects to, we just had the CPO of Uber, the current CPO of Uber on the podcast, and he had a few episodes before this one. It was all about dog fooding and basically exactly discovering these problems. He’s done seven to 800 rides as an Uber driver to discover these problems. He had this great quote about, it’s one thing to watch, just build an app for drivers sitting in your office making it look really pretty. It’s another to be driving 60 miles an hour with this phone a few feet away from you trying to figure things out.
Peter Deng: A hundred percent. Oh, I remember that I took two weeks off before I joined Uber. And in that time I’ve been obsessed with user research for the longest of times, and this is more relevant back then when you wanted to really understand how the wide massive users were using your product. And I remember I actually leased a car to drive for Uber those two weeks. So it was a little white VW something or another. I put an Uber sticker on it, I turned on the app and it just started driving and there’s no better way to learn than to dog food, and I’ll just build on what… Sachin is the person you had on the podcast? Yeah, he’s an amazing, amazing guy. And so I’ll just build on what he said there. I think that what really stuck with me in terms of framework that I learned back in school because I was brought up with the IDEO way of design thinking and I was at the design school at Stanford where before we literally were in trailers. That’s how early it was.
But I remember the framework that really stuck with me is what IDEO preached, which is there are five stages to great design thinking. Number one is empathize, two is to define, three is to ideate, four is a prototype, and five is to test. And what I love about this framework, and I really hope this doesn’t get lost because I don’t know how much it’s being preached nowadays in design thinking is that it has the right words associated with it. The first thing is empathizing. You’ve got to really feel the pain of your customers. It’s not just about theoretically understanding what the problems are. It’s really empathizing, which is why user research was so important to me is to understand that, or even like Sachin said, just taking those rides but also flying around the world. And when I was working at Uber to figure out, well, what are the various conditions?
And so empathize is a really powerful word. The define is also a really powerful word because it forces you to articulate what the problem is. And this is, again, going back to the language thing of you have to be very intentional about defining the problems that you want to solve and then ideate, we all know it’s brainstorming and prototyping and tests are self-explanatory, but the first two stages I think are really insightful and it talks directly to what Sachin was saying. You’ve got to dog food because you really have to empathize and the great products are when you really feel the pain and you really empathize with what people are experiencing.
Lenny Rachitsky: That’s a great connection to another podcast episode that came to mind as you were talking, the head of product at Linear, Nan, had this really great concept that’s exactly what you’re saying, which is as a product person, you want to feel the pain of your customer the same way they do. You shouldn’t stop asking questions to understand what they’re telling you until you feel the pain that they feel and that’ll help you. Basically, that’s like how to operationalize empathizing. It’s just do you feel the suffering?
Peter Deng: Yeah, and I really do hope product people still do this to this day because I think there’s so many shortcuts that if people take, you’re going to miss the point, right? I still remember distinctly flying down to LA with Kevin Systrom to go do a user research study, and it was a one-way glass thing where we listened to people talk about Instagram and how they use Instagram, and there’s no substitute for that. I think that to anyone out there who’s doing user interviews and then saying, hey ChatGPT, summarize the takeaways, you’re missing the point. You can’t empathize with the summary. You have to be in the room fully immersed, no phones, just actually hearing the words and the intonation. That’s how you’re going to get the full color.
Lenny Rachitsky: It makes me think Jeff Bezos has this great quote, if you have an anecdote and data and they’re telling you different things, trust the anecdote. Oh, man. So many lessons. Okay, so to start to kind of wrap up our conversation, we covered a lot of ground, I want to ask you about Facebook real quick. So you joined Facebook very early. Eric Antonow, who I’ve mentioned previously, told me that it was very strange that you left Google to join Facebook at that stage. Google was killing it, on top of the world. You had such a strong career path, things were going great, but you decided to take a big leap joining Facebook. What did you see? Because I think there’s something interesting here that we can learn about what you saw that may help other people decide where to go work.
Peter Deng: I’ve always been enamored with this idea of understanding us as fundamentally human and how we’re wired. And I remember at the time talking to the folks at Facebook and seeing it, and this was back when people were like, oh, this is just a college site, and that was the vibe back then. But what I saw was that the team and Mark and others really understood the fundamental human desires that people had to connect and feel lonely and to share, and they really got the right articulation of the problem they were trying to solve, which was to make the world more open and connected. And this really resonated with me because again, I study a lot in college like psychology, and I was really enamored with this idea of how are we as humans fundamentally wired? And it felt to me like a no-brainer to go work at Facebook because they saw how people were wired and how to actually build products that complement how people are wired.
And it wasn’t that they were trying to force fit something into something that was unnatural. It was almost like how do we build technologies and products that actually augment our fundamental desire to stay connected? And this goes back to why I think the power of wars is so important is because you take a look at some of the mission statements for Friendster or MySpace, I don’t even know if they had mission statements or what they were, they were kind of vapid and they didn’t really speak to the fundamental humanity of what Facebook was striving to build and that just deeply resonated with me. And so I remember spending time with Eric being like, “Hey, what should I do? Should I take this offer from Facebook or should I stay at Google?” But ultimately it was just that deep resonance with my values of building things that were fundamentally human. And ultimately I think that for any startup out there, anyone building product, the more that you can get a good impedance match between what you’re building and what humans fundamentally want and need, the more successful you’re going to be.
So that’s my big answer. I think the secondary answer, I’ve always optimized for learning in my career, and this is a huge thing that I say to a lot of people because they look at sort of like, oh, you’ve been at all these companies, what’s your secret? I’m like, well, I’ve just figured out that I want to go to the place where I can learn the most. And for me, that wasn’t really Google, but I had so much I wanted to learn from operating at Facebook. And at Facebook I would say, yeah, I was there for nine and a half years, but I always jumped around every two and a half or so when I feel like there was something new to learn. And that’s it.
I mean, I don’t know if it’s a secret or not, I got lucky and I was able to have opportunities to learn different things and different skills, and that served me quite well. And regardless of any outcome, I would say that’s just a great way to live your life personally is just to optimize for learning and those experiences and for me, moving to Facebook was that I saw so much learning that could have happened and it ultimately did happen. So I feel like that was a good outcome too.
Lenny Rachitsky: [inaudible 01:39:55] did it. So a couple takeaways here for folks that are maybe trying to decide between a couple roles, maybe deciding if they should leave and do something new is one, are you feeling like you’re learning enough/is the new place you’re thinking about going to help you learn a lot more? Two, is what they’re building aligned with human behavior? Almost this impedance match that you described. It feels like there’s another element you shared, which is do they have a really unique insight about how things work? And also do you really care about this? Is this also how you see the world? So you’re talking about a Facebook, they have this really unique insight about human behavior and that was really important to you, and so it was a really good fit.
Peter Deng: A hundred percent. Yeah. I think the insight thing, thank you for summarizing that and drawing that out because it’s also what I look for and what I want to partner with companies and startups now is do you have that unique insight? Are you teaching me something that I really don’t know? And that usually is a good indicator of a strong point of view, and having a strong point of view is really important because there’s a saying that Mike and Kevin had at Instagram which is we may not be right, but at least we’re not confused. I think it’s a beautiful phrase I thought because sometimes you’ve just got to go and do the thing that you think is right and the indecision is going to be one of the things that really gets you and bites you. So that for me is something as I look for folks who have a strong conviction, whether it’s the founders I support when I go join and be an operator at the company or the founders I support in my current role.
Lenny Rachitsky: That’s so interesting. Tomer Cohen, the CPO of LinkedIn, that’s a famous phrase that he often uses too.
Peter Deng: Really?
Lenny Rachitsky: So I think he borrowed it from those guys. Yeah. That was one of his mottos. We may not be right, but we’re not confused.
Peter Deng: Wow, I didn’t know that. So I did talk him at one point. I don’t remember if that’s something we talked about, but again, it could just be like great minds think alike, and we just had different great folks with Mike and Kevin and Tomer feeling the same vibes.
Lenny Rachitsky: I love just how many episodes this conversation has referenced. Okay, so speaking of learning, final question before we get to our very exciting lightning round, I’m going to take us to Fail Corner, which is very aligned with your growth mindset question. So the idea of this segment is people come on this podcast, they share all these amazing stories of everything’s working out, I had so much success, worked at all these incredible companies, everything worked, but in reality, things don’t often work out. Most people go through a lot of failed initiatives, projects, career hits, so the question is just what’s a product that you built and launched that was just a big failure? And I’ll ask it the way you ask it, how did that change the way you think and operate?
Peter Deng: One example is, since we were talking about Instagram before, we tried to build a kind of camera first app at Instagram. It was called Bolt and it didn’t work and the great levels of craft and design and the premise was essentially can we make it so it reduces the pressure to share, and you can open to a camera, you can just send some things to folks and you get some good feedback and you go from there. And it was obviously the Instagram design team, so it was top-notch. The app was designed really well. It was really fast because it’s the Instagram engineering team and they were just really good at making performant mobile apps. It had all of the advantages that we had talked about that we valued at Instagram, but we launched it and I believe it was New Zealand or Australia and it didn’t work.
And I remember the reason we knew this is as we were looking at sort of the retention graphs and retention is the key indicator in any product that you build, it’s not the number of users, not the volume, it’s actually retention and cohorted retention, you can [inaudible 01:44:00] the line and if it asymptotes, then you’re in a good spot because that means that people over X period of time will continue to stay on the app and that just didn’t happen. And I think the learning here was that you can really have the best team in the world with the best product taste and you can’t really predict what’s going to hit on the first go.
And failure is okay, you’re just going to up and learn from that and nobody wallowed over that. We actually had some technology that we built there that we were able to port over to the main app, which was really helpful, but to quote the great american poet Sean Carter, “It ain’t a loss, it’s a lesson.” And I think it’s really important that you see that as a product person is that you don’t see it as failure, you see it as kind of great. Now I’m that much smarter. And this is something that I’ve just collected. There’s other examples as well, but I think this is a good example of sort of something that’s somewhat counterintuitive, that you have the best team, you’re going to provide those hits over and over, but sometimes you can’t predict those hits and you just have to have the wisdom to be like, okay, let’s see what we can learn here, see what we can save here, and then move on.
Lenny Rachitsky: I absolutely remember that product in launch or heard about it, but I also don’t ever think about it. And so I think it’s a good reminder. Because Instagram launching a new product that’s trying to rethink the way you do your camera, that’s a big deal. And so I could see that being a really big deal for it not to work out. At the same time, nobody remembers that really.
Peter Deng: Exactly. Yeah.
Lenny Rachitsky: Peter, we’ve gone for two hours at this point. I feel like we could do two hours more. We’ll save that for another conversation.
Peter Deng: Great.
Lenny Rachitsky: Before we get to our very exciting lightning round, is there anything else you either wanted to share or want to leave listeners with to maybe double down on a point you made that you think might be helpful? Otherwise, we’ll just jump right in.
Peter Deng: I think we should jump right in because I feel like you’ve extracted every little ounce of what wisdom I had here and you did a great job here just helping me remember these stories and recounting stuff, so I’m ready to jump in.
Lenny Rachitsky: That’s my goal, although I know there is much more that I haven’t even started to tap, but with that, we reached our very exciting lightning round. Are you ready?
Peter Deng: I’m ready.
Lenny Rachitsky: Question one. What are two or three books that you find yourself recommending most to other people?
Peter Deng: This is easy for me. Number one is Sapiens. If you’re a product person, you have to understand our own humanity if you want to build products for people, straight up. That’s a beautiful book. I read it before it was called Sapiens, it was called From Animals to Gods, and it was just republished in a different name, but it has really stuck with me and I remember, it’s a very short, easy read, so I’d recommend that. The second book I think for product folks is a classic one, which is The Design of Everyday Things by Don Norman. This may seem outdated and old, but I promise you it’s not. It really helps you understand physical product design, which is again, things that mold and shape to humanity. I think it gives you a good sense of that.
Third book is something I’m reading right now it was recommended by a friend of mine and I can’t put it down. It’s called The Silk Roads by Peter Frankopan. And basically this is a recounting of history through the lens of The Silk Road and the Middle East and how that’s evolved. It’s so fascinating because one of the things I love, Lenny, is seeing things from different perspectives. This is why travel’s fun, this is why user research is fun for me, and it really helps you see the events of world history that we’ve all been experiencing through a very western viewpoint in a different way. And it kind of connects a bunch of things that are like, there’s Western thought, there’s Eastern thought, but if you see the connection between them, it’s super fascinating. I’m only two, three or maybe four chapters in, but definitely something I would recommend off the bat.
Lenny Rachitsky: What is a favorite recent movie or TV show that you’ve really enjoyed?
Peter Deng: Maybe it’s not as recent, but the one that always comes back to me is The Wire, HBO’s The Wire. And I guess there’s just so many TV shows now that I’m still processing, do I want to put it in my all-time greats? But the storytelling there and the various different sort of consistent characters, but the fact that there’s the beautiful writing of The Wire is something that’s unparalleled.
Lenny Rachitsky: I’m now curious what’s in your all-time greats list, but I’m not going to go there. We’re going to keep going. What’s a favorite product you’ve recently discovered that you really love?
Peter Deng: I’m just going to go with Granola because I think that we talked about this before, but this has been a superpower for me and I have a lot of commute time now. What I do is I just do a single player mode. I go up and I start thinking about and brainstorming about sort of ideas or theses I have for investing or whatnot, and I get to where I’m going and boom, they’re organized in a more cogent way and oftentimes ways that I didn’t even think about articulating them. So it goes through the process of forming words, but it also helps with that assistance and I think it’s a beautiful product on many different levels.
Lenny Rachitsky: Wow. Granola’s killing it at this category recently, and I’ll give a shout-out, you get a year free of Granola if you become a yearly subscriber of my newsletter, which is not just for you, but your entire team, they gave an incredible deal.
Peter Deng: Is that true? I didn’t know that.
Lenny Rachitsky: A hundred percent true.
Peter Deng: Okay, well I’ll tell you, I was not compensated for that little pitch there, that’s genuine right there.
Lenny Rachitsky: I’m also not compensated. Yeah. If you go to lennysnewsletter.com and click bundle, you’ll see a way to get it. Love the product, use it all the time. I should be using it for these interviews and then I could have a whole summary ready to go. Okay, next question. Do you have a favorite life motto that you often come back to in work or in life?
Peter Deng: Yes. This is actually something that my dad taught me. It’s a saying that is in Chinese. It actually rhymes in Chinese but kind of almost rhymes in English. And it goes something like this in English which is if you move a tree, it dies, but if you move a person, he thrives. And I think it’s a really interesting thing I keep on coming back to, and this goes back to why for me it’s just the joy of learning and trying new experiences and being at different companies that I’ve been very fortunate to be at. I really think that that’s how you should live life is just to kind of experience these different experiences. And it’s kind of poetic to be like, yeah, unfortunately for trees, you can’t really move them after a while. But for humans, I think that you move them around and we get different travel experiences and we get different life experiences when we go to different jobs, and I think that makes life really worth living.
Lenny Rachitsky: I always think about what I would answer to this question, and there’s a few, but one is something I always come back to when my wife and I are deciding to do something is choose adventure. Similar sentiment. Final question. So you’ve now moved from product leader to investor, so I just want to give you a chance to tell people what kind of stuff you’re looking for. So you moved [inaudible 01:51:11] now, investing in startups. What sort of startups are you looking for? Who should reach out if they’re interested in-
Peter Deng: Well, I appreciate that opportunity. Look, for me, I think it’s been very clear. I just love working with great people and for me, investing is just the ability to support more amazing founders. I’ve always been drawn to the founder archetype, like working closely with Zach or with Travis or Howie, Brendan at Oculus, and folks at Opening Eye, I think there’s this amazing sort of visionary person that I love supporting in one way or another. And I’ve supported them mainly from the inside as a product leader, but for me it’s just finding those amazing founders. In this current role, I get to work with many founders at the same time. And just two days ago I had meaningful calls, product jams with three different founders in three different industries, and that kind of keeps my mind super alive. So that’s kind of why I’m doing what I’m doing now, and I would love to find some more of those amazing thought partners and people that I can just help out if I can.
Lenny Rachitsky: Okay. Stage and market, anything there for folks of like, okay, he’s a fit, not a fit.
Peter Deng: Absolutely. So I would say early stage seed, seed plus and A is where I really get excited. I feel like I am able to help folks see the next stage. I’ve seen a lot of movies in my life in my career, so it’s like, oh, great, I can definitely see this extrapolating out. You’d have to convince me of the future, and then it’s really fun to be able to jam and help support if I can in how you scale from the one to 10 and 10 to a hundred. So that’s really big.
And then in terms of what I look for it’s the two things I said before, in this day and age, there’s so many amazing things that’s going to be built. One is do you have unique data and do you have a data flywheel? Two, do you have a really crafted workflow that you can really get after? And I guess third, do you have that insight of what product things actually matter and also which ones don’t? And then how do you actually go and expand upon that? So yeah, really excited to meet a bunch more founders, whether it comes from here or somewhere else.
Lenny Rachitsky: Okay, so final question and it’s how do folks reach out if they want to actually talk to you about this and how can listeners be useful to you?
Peter Deng: Thank you for the question. I am an introvert, so I’m really kind of silent on a lot of social media. I have accounts on X and Threads, but really I think LinkedIn is the network of choice for me. I want to be able to passively consume and learn about what’s happening. How listeners can be helpful, I just want to learn. What are you all thinking about? What are some of the insights you’re seeing? One of the analogies I have about AI in this day and age is that it’s this really interesting new element that humanity has discovered. And what’s awesome is that humanity is also very creative. And so what humanity does with this new element, I’m fascinated by, and you can tell the founders who’ve actually played with this element because they have this innate sense of what this thing can do and can’t do, and I’m just looking to be inspired by the creativity of all you all out there.
Lenny Rachitsky: Wow, that’s such a cool way of thinking about it. It’s going to change my perspective on AI a little bit. Peter, this was incredible. I really appreciate you taking the time to share so much wisdom. I know this is the first time you’ve done anything like this. I feel like this is going to help a lot of people in a lot of different ways. I feel like we covered everything I wanted to cover, so just again, thank you for-
Peter Deng: Well, thank you for having me. This has been a real pleasure and hopefully some folks out there can get some learnings from this and find it useful, but that was my goal is to be able to share some things and hopefully it’ll be helpful to some folks out there. So thank you. Thank you for the opportunity.
Lenny Rachitsky: Thank you, Peter. Bye everyone. Thank you so much for listening. If you found this valuable, you can subscribe to the show on Apple Podcasts, Spotify, or your favorite podcast app. Also, please consider giving us a rating or leaving a review as that really helps other listeners find the podcast. You can find all past episodes or learn more about the show at lennyspodcast.com. See you in the next episode.
Glossary
| English | 中文 |
|---|---|
| AGI | AGI(通用人工智能,保留原文缩写) |
| Andrew Chen | Andrew Chen(人名,Uber 增长团队负责人,保留原文) |
| Bolt | Bolt(Instagram 曾推出的相机优先应用,保留原文) |
| cohorted retention | 分群留存(cohorted retention) |
| Copilot | Copilot(微软 GitHub 的 AI 编程助手,保留原文) |
| CPO | CPO(首席产品官,Chief Product Officer,保留原文缩写) |
| Cursor | Cursor(AI 代码编辑器产品名,保留原文) |
| data flywheel | 数据飞轮(data flywheel) |
| design thinking | 设计思维(design thinking) |
| dog food | 内部试用(dog food,指公司员工在产品公开发布前内部使用) |
| Don Norman | Don Norman(人名,保留原文) |
| Dwarkesh | Dwarkesh(人名,播客主持人 Dwarkesh Patel,保留原文) |
| Eric Antonell | Eric Antonell(人名,保留原文) |
| Eric Antonow | Eric Antonow(人名,保留原文;注:前文术语表中为 Eric Antonell,此处保留原文写法) |
| ETA | 预计到达时间(ETA,Estimated Time of Arrival) |
| Fail Corner | 失败角(Fail Corner,播客常设环节) |
| Felicis | Felicis(风投机构名,保留原文) |
| Fidji | Fidji(人名,指 Fidji Simo,保留原文) |
| Friendster | Friendster(社交网站名,保留原文) |
| George Lee | George Lee(人名,Instagram 增长团队负责人,保留原文) |
| Granola | Granola(AI 会议笔记产品名,保留原文) |
| grit | 毅力(grit) |
| growth mindset | 成长型思维(growth mindset) |
| growth PM | 增长产品经理(growth PM) |
| Herbert Clark | Herbert Clark(人名,斯坦福大学心理学教授,保留原文) |
| IDEO | IDEO(设计咨询公司名,保留原文) |
| Ilya | Ilya(人名,指 Ilya Sutskever,保留原文) |
| impedance match | 阻抗匹配(impedance match,借用电子学术语,指产品与用户需求的契合度) |
| Jeff Bezos | 杰夫·贝佐斯(国际知名人物,使用公认中文译名) |
| Jill | Jill(人名,Uber 公关/传播/政策负责人,保留原文) |
| Joanne Jang | Joanne Jang(人名,OpenAI 员工,保留原文) |
| Josh Constine | Josh Constine(人名,保留原文) |
| Kevin | Kevin(指 Kevin Systrom,Instagram 联合创始人,保留原文) |
| Lauryn | Lauryn(人名,Airtable 增长负责人,后转至 Notion,保留原文) |
| Lenny Rachitsky | Lenny Rachitsky(主持人名,保留原文) |
| Linear | Linear(项目管理工具产品名,保留原文) |
| LinkedIn(职场社交平台产品名,保留原文) | |
| LLM | LLM(大型语言模型,Large Language Model,保留原文缩写) |
| managing up | 向上管理(managing up) |
| Marc Benioff | Marc Benioff(人名,Salesforce CEO,保留原文) |
| Mark | Mark(指 Mark Zuckerberg,Facebook 创始人,保留原文) |
| Mike | Mike(指 Mike Krieger,Instagram 联合创始人,保留原文) |
| model designer | 模型设计师(model designer) |
| MVP | MVP(最小可行产品,Minimum Viable Product,保留原文缩写) |
| MySpace | MySpace(社交网站名,保留原文) |
| Nan | Nan(人名,Linear 产品负责人,保留原文) |
| newsfeed | 动态消息(newsfeed) |
| Notion | Notion(产品名,保留原文) |
| NPR | NPR(美国国家公共广播电台,National Public Radio,保留原文缩写) |
| Peter Deng | Peter Deng(人名,保留原文) |
| Peter Frankopan | Peter Frankopan(人名,保留原文) |
| photo tagging | 照片标注(photo tagging) |
| post training | 训练后(post training,AI 模型训练后的微调阶段) |
| PRD | PRD(产品需求文档,Product Requirements Document,保留原文缩写) |
| product craft | 产品工艺(product craft) |
| product taste | 产品品味(product taste) |
| PXD | PXD(Peter Deng 的个人工作风格文档,保留原文) |
| QBasic | QBasic(编程语言,保留原文) |
| Replit | Replit(在线编程平台产品名,保留原文) |
| RPG | RPG(角色扮演游戏,Role-Playing Game,保留原文缩写) |
| Sachin | Sachin(人名,指 Sachin Kansal,Uber 现任 CPO,保留原文) |
| Sean Carter | Sean Carter(人名,即 Jay-Z,保留原文) |
| single player mode | 单人模式(single player mode) |
| South by Southwest | 西南偏南(South by Southwest) |
| The Wire | 《The Wire》(火线,HBO 电视剧名,保留原文) |
| Tomer Cohen | Tomer Cohen(人名,LinkedIn CPO,保留原文) |
| Uber Reserve | Uber Reserve(Uber 预约出行产品名,保留原文) |
| zero shot | 零样本(zero shot,AI 模型在无示例情况下直接执行任务) |
Reformatted by reformat_english.py
Peter Deng(Meta、Instagram、Uber 产品负责人)
文字稿
开场快问快答
Lenny Rachitsky: 你打造并领导了 Facebook 的 Newsfeed,将 Messenger 作为独立应用推出,还发布了 ChatGPT Enterprise。关于把一件事从创意做到十亿级规模,你学到的重要一课是什么?
Peter Deng: 你必须提前规划好棋步。你必须真正在行动之前深思熟虑,并构建能让你持续高速前进的系统。
Lenny Rachitsky: 你学到的最反直觉的一课是什么?
Peter Deng: 有时候你的产品本身其实并不重要。在 Uber 我学到了这一点——在 Uber,价格和预计到达时间才是真正的产品。从全局视角来看,我们人类消费的是产品的整体。这不是说你不应该修 bug,但它的影响力远不如那些对人们更重要的东西。
Lenny Rachitsky: 你认为 AI 会带来什么重大变化,是人们还没有充分意识到的?
Peter Deng: 教育将会改变。我儿子当时九岁,自己搭建了一个自定义 GPT——你输入任何主题,它就能给你一句包含英文字母表中每个字母的句子。这难道不令人惊叹吗?我已经能看到他的大脑在重新连线。
Lenny Rachitsky: 你在招聘时看重的一点是什么?
Peter Deng: 六个月后,如果我还在告诉你该做什么,那我就招错人了。这能让我和对方在一个更高的层面协作——目标不是”你有没有完成这个 OKR”,更高层面的目标是:我们的校准够不够?我们是否真正达到了这样一种状态——六个月后是你来告诉我需要做什么?
Lenny Rachitsky: 关于如何成为一名优秀的产品人,你学到了什么?
Peter Deng: 我认为产品经理有五种不同类型。第一种是——
嘉宾介绍
Lenny Rachitsky: 今天的嘉宾是 Peter Deng。Peter 也许是你从未听说过的、最具影响力的产品领袖。我常说,最优秀的产品人不是那些在 Twitter 和 LinkedIn 上分享建议的人,而是那些没时间做这些的人,因为他们太忙于做实际工作了。Peter 就是最好的例子。他曾是 OpenAI 的产品副总裁,负责 ChatGPT 的产品设计和工程,协助推出了 ChatGPT Enterprise、语音、记忆、桌面端、自定义 GPT 等功能。他还负责搭建了他们的第一个增长团队。
他是 Instagram 的首任产品负责人,与 Mike 和 Kevin 密切合作,负责所有产品开发,包括内容分享、广告、增长,甚至帮助建立了设计和用户研究团队。他也是 Uber 乘客产品团队的负责人,负责乘客 App 中的所有内容,包括对 Uber Pool 和机场上下车体验的重大改进。他还帮助团队推出了新产品,包括 Uber Reserve,如今这已是一项年营收接近 50 亿美元的业务。
他还在 Facebook 工作了近十年,是公司第四位产品经理,打造并领导了当前 Newsfeed 产品背后的团队、独立的 Messenger 应用,还包括照片、小组、主页和个人主页。他还曾是 Airtable 的首席产品官,帮助公司系统化产品构建方式并向企业级转型。他还领导过 Oculus 的产品管理。如今他是 Felicis 的普通合伙人,能够以投资者的身份将自己所学的一切带给更多创始人。他此前从未做过播客,也从未公开分享过这些经验或故事。所以,你将大饱耳福。
感谢 Eric Antonow、Nick Turley、Lauren Motomati、Joanne Jain 和 Sundeep Jain 贡献的问题和话题。如果你喜欢这档播客,别忘了在你最喜欢的播客应用或 YouTube 上订阅关注。此外,如果你成为我通讯的年度订阅者,可以免费获得一年大量优秀产品的使用权,包括 Bolt、Linear、Superhuman、Notion、Perplexity 和 Granola。请访问 lennysnewsletter.com 点击 bundle 查看详情。话不多说,请出 Peter Deng。
Lenny Rachitsky: Peter,非常感谢你来到这里,欢迎来到播客。
Peter Deng: 谢谢。我非常激动能来这里,真的很荣幸。期待度过一段美好的时光。
Lenny Rachitsky: 在我们为这次对话做准备的时候,我们在讨论应该聚焦什么内容。我们今天要聊的东西很多。但你说过一句话特别有意思,我很想从这里开始——你说你一直觉得自己无法说出所有真正想表达的想法和感受,因为你身在企业内部,公关人员让你保持统一口径,而这是你第一次感到可以自由分享。
Peter Deng: 第一次。
Lenny Rachitsky: 好,那么首先,这种感觉如何?其次,跟我们说一件你一直想分享的、或者终于可以公开谈论的事。
Peter Deng: 感觉真的很好。所以,让我先说……我很高兴你一上来就抛了个犀利的问题,让我先补充一些背景。我今天确实可以更自由地发言,但也并不完全是大家想象的那样。我不是来泄露任何公司机密的。但我天生是个爱讲故事的人,某种程度上也算是个内向者。所以这期播客,我觉得可以跟你就任何话题深入聊下去,把一些背景补充完整。因为我认为,缺少了一些背景的话,我的一些犀利观点之类的可能会被断章取义。而且不用赶时间、不用觉得有一条公关口径必须去踩,这种感觉真的很自由。所以,棒极了,你心里觉得对听众有趣的任何话题,我都在这儿,嗯,我很兴奋。
Lenny Rachitsky: 我经常跟嘉宾说一件事,我也不希望这被人断章取义——我总是把自己形容为一个”反向记者”,我希望嘉宾展现出自己最好的一面。我绝不想让人措手不及,或者说出他们本不想说的话。所以——
Peter Deng: 太好了。
Lenny Rachitsky: 这里是安全空间。好吧,但还是想问,有没有什么你想分享的、可能很有意思、你一直想分享但之前没法说的?有没有这类的东西?
Peter Deng: 嗯,我经常被问到的一个问题大概就是,AGI 会不会来?会不会解决一切?
Lenny Rachitsky: 你看到了什么?
Peter Deng: 嗯,这很有意思,因为我在 OpenAI 的时候,正值人们真的对 AI 很恐惧的阶段——“天哪,它要消灭人类了”或者”它要做所有这些事了”。但我觉得每一项技术,大家都需要一段时间去适应。AGI 也是类似的情况,它离我们还很远,所以每个人都在想:“我们的世界到时候会是什么样?“而真正的答案是,我们谁也不知道。但就解决问题而言,我觉得有些人认为 AGI 会解决一切,但我不这么认为。AGI 只是必要条件,而非充分条件。大量的价值仍然需要许许多多建设者的努力打拼,才能真正把这种新的能源转化成我们人类愿意使用的、能解决我们问题的东西。那种努力是必不可少的,那种苦功夫是必不可少的,才能真正让 AGI 变成有用的东西。
Lenny Rachitsky: 你的意思是,人们觉得 AGI 一到,突然间所有工作都没了,AGI 会做一切。因为我觉得这其实是一个乐观的信息——事情会没事的——当 AGI,基本上 AGI 的意思,我好奇你有没有一个明确的定义,但 AGI 基本上就是 AI 跟人类一样聪明——
Peter Deng: 嗯,我不——
Lenny Rachitsky: 一般意义上的。
Peter Deng: 我不敢自称这方面的专家,但我认为每一项新技术出来,我们都能驾驭它,而且这种驾驭需要付出很多努力。“驾驭”这个词我是非常刻意使用的。我举一个非常基础的例子。今天看起来理所当然的事情是,曾经有一段时间数据库风靡一时。就像,“天哪,你可以存储大量数据,而且可以非常快地查询,想象一下各种可能性。“我认为很多出色的企业家和建设者在数据库之上打造了一些非常棒的产品。
Lenny Rachitsky: 没错。
Peter Deng: 事实上,这正是我们今天看到的所有东西的基础。今天看来这似乎不言而喻,但我不知道,也许十年、十五年后我们回头看的时候,会说,“当然啦,有这么一个超级智能的思维机器,这理所当然嘛。“但它需要产品建设者走进去说:“我们如何把这种能量引导成我们人类喜欢用、愿意用的东西?”
Lenny Rachitsky: 我很喜欢这种乐观主义。就是——当计算机跟人类一样具备通用智能的时候,事情不会失控。
Peter Deng: 我想这正是我想说的。而且我认为,每一项技术人们都会有这种恐惧。我记得有一次看一部纪录片,他们讲到自行车刚出来的时候,人们就说,“天哪,这会是万物的终结。“同样,今天听起来很荒唐。因为你会想,“自行车?真的假的?“但如果你把自己放在那个时代、那一代人的心态中——而下一代人回头看这期播客时也会像看上一代人一样——我认为,同样地,我持乐观态度,事情会没事的,我们会适应的。实际上这也是我在西南偏南(South by Southwest)上跟我的朋友 Josh Constine 聊到的一点,就是人类始终会与技术共同演化的观念。我认为这种共同演化已经在发生了。
如果你去看看,ChatGPT 刚出来的时候确实有很多人对 AI 感到恐惧,但当你开始熟悉它之后,情况就会发生变化,你就能够从恐惧进化到熟悉,再到完全驾驭这个东西——“天哪,看看现在涌现的所有创业公司,我们能构建的所有东西,仅仅 18 个月。“我想说,我们回头看,大家的态度已经发生了转变。所以我猜我的乐观有一部分来自于:如果你回头看 18 个月,再向前看 18 个月,我们现在追逐的某些东西,会不会也是同样的情况?
AI 将如何改变教育
Lenny Rachitsky: 好,让我沿着 AI 这条线再多追问一点,然后我们可以转到其他话题。我感觉每次对话都有一个聊 AI 的时段,然后就觉得,好吧,还有其他同样重要的事情。那么让我问你这个问题——你认为 AI 会以什么方式发生巨大改变、但人们还没有充分意识到的一件具体的事是什么?
Peter Deng: 我认为教育会发生巨大改变。我经常思考这个问题,因为我相当深入地参与了孩子学校的事务,这是我离开 OpenAI 之后做的事情。让我着迷的是,看着我的儿子——他在公开之前就内部试用(dog food)了不少 OpenAI 的东西,我想我可以安全地说这点,应该没问题。他玩 ChatGPT 和一些最新模型的时候才九岁,我已经能看到他的大脑在重新布线。他开始提出各种问题,他从来没听过”prompt”这个词,但是——这就体现了人类心智的了不起之处——因为他在幼年就接触到了这项技术,有些东西自然而然就被解锁了。我认为你能够用不同的方式思考。我给你举一个具体的例子。
他去上 Python 课,在写代码。实际上,我不认为他长大后还需要写代码。我觉得那会是一个已经被解决的问题。但这是一项非常有价值的技能,因为我认为学编程就是学习如何以结构化、系统化的方式思考。而他给 ChatGPT 输入了一些我根本想不到的非常疯狂的东西。其中一个是:“嘿,ChatGPT,你能给我一个包含字母表中每一个字母的句子吗?主题是海洋,或者主题是太空?“
编程与抽象思维
Peter Deng: 这之所以让我震撼,是因为在传统编程中你写不出那个程序。你不能在 Python 里说”哦,写一个函数来生成这样的句子”。这真的是一个非常难写的函数。但他能想到那样的提示词,真的很酷,因为他做了一个自定义 GPT——你输入任何主题,它就会给你一个包含英文字母表中每个字母的句子,有点像”the quick brown fox jumped over the lazy dog”。这不是很令人惊叹吗?九岁的年纪就能想到这些,而我九岁的时候还在玩乐高,顶多玩玩 QBasic。所以,年幼人类的大脑会因为这项新工具而发生怎样的演变,这种认知将改变我对教育方式的看法。坦白说,我不是教育专家,但我确实对此思考了很多。我认为未来非常重要的一点,是学会如何提出正确的问题。人类天生就充满好奇心。但把好奇心转化为向 AI 提出的正确问题——毕竟 AI 是所有人都能接触到的——这将决定你能做出什么样的工作,成为一种关键的差异化能力。
我打个比方:计算器发明之后,人们并没有停止做数学,他们只是开始做更高层次的数学。计算器解放了大脑,让人去做其他事情,在更高的抽象层次上思考。我认为我们必须让孩子们准备好去思考:“你怎么在更高的抽象层次上思考?“这种事以前也发生过。Google 让记忆在某种程度上变得过时了——你不再需要背诵事实,搜一下就行。下一个阶段将是类似这样的:“代码只要你召唤就会出现。“那么人们将会思考什么?我们需要培养哪些处于更高抽象层次的技能,那些能够激发创造力、激发好奇心的技能?这将非常有意思。所以我认为教育会发生巨大的改变,就像过去进步主义教育从背诵乘法表转向更高层次的思维训练一样。我认为这将是发生重大变革的领域之一。
Lenny Rachitsky: 这让我想起最近在听的一篇 NPR 报道,他们跟踪了几位用 ChatGPT 设计课程大纲的教授。之前有很多关于学生用 ChatGPT 作弊、让 ChatGPT 代写论文的讨论。但老师们也在大量使用 ChatGPT。然后学生们反过来严厉批评教授,因为他们发现教授在用 ChatGPT 设计课程。这变成了一种军备竞赛。
Peter Deng: 但这也很有意思,因为事情还在进一步发展。整个系统都需要改变。因为我依然相信人类大脑天生就是好奇的,我们仍然需要某种方式的发展。但发展的方式会是怎样的,我很着迷地想看看这一切如何展开。
语言的力量
Lenny Rachitsky: 我想回到产品话题上,但在此之前,我知道你在这方面有很多思考。我在和你共事过的人那里多次听到这一点——你非常强调语言的力量和重要性,认为在写作和口语中认真推敲用词至关重要。能谈谈你对此的看法吗,就是作为领导者,语言的重要性和力量。
Peter Deng: 我记得大学时上了一门让我印象深刻的课,叫”语言与思维”(Language and Thought),由 Herbert Clark 教授。他有一个让我非常震撼的论点,就是”语言实际上会影响你的思维方式”。这是他论点的一部分。自从听到这个观点、在他的书里读到它、听完那些讲座之后,我就再也无法停止思考这个问题,因为它太准确了。我从小说中文,我觉得中文里有很多东西,让我在学了英语之后注意到自己思维方式上的差异。这方面也有一些研究,我记得——具体我不太确定,需要去查证——但我记得在俄语中有两个不同的词表示蓝色,一个是偏绿的蓝,一个是亮蓝之类的。
Lenny Rachitsky: 我会说俄语,不过我六岁就搬到美国了,所以俄语不太好。你说的这些让我在努力回想,你继续。
Peter Deng: 好,那太好了,我需要找个方式验证一下。但据我记忆,因为俄语有这两个不同的词来表示这两种不同深浅的蓝色,说俄语的人在学习英语之后,区分这两种蓝色比只说英语的人更容易、更快。我读到过一些相关研究。还有一些语言据说根本没有表示蓝色的词,这导致他们在长期中很难区分这些颜色。这些对我影响很深,我觉得确实很有道理。所以在实践中,当我制作幻灯片的时候——几周前我给一个班做了一次演示,整个幻灯片上总共大概只有 20 个词。但我花了几个小时反复推敲每一个词,因为我真的想确保准确捕捉到我要表达的核心意思。我认为在产品工作中,这种打磨非常重要。因为当你坐下来写一份愿景文档或 PRD 时,如果你不在意自己用的词,不刻意选择措辞,这些都会产生下游影响。人们可能会误读,某些隐含意味可能无法传达。所以我非常谨慎,因为我认为用错一个词会有倍增效应和连锁反应。我深信语言影响思维这个论点,正因如此我一直对此格外用心。
Lenny Rachitsky: 嗯,对。而且我觉得 AI 也能在这方面帮助你。
Peter Deng: 没错,完全正确。
Lenny Rachitsky: 我们做过一期——
Peter Deng: 其实说到 AI,这真的是一个很有意思的点。我觉得人工智力的突破来自大型语言模型(LLM),这件事既有趣,又富有诗意、恰如其分。让我觉得很有意思的是,每一个词、每一句话中都凝聚和塑造了大量的知识。当 ChatGPT 做出一些真正有趣的事情时,我告诉人们,它往往只是在写 Python 代码并加以解释。而 Python 又是一种语言。所以我觉得这里面有一个很有意思的关联——人类思维在语言中的凝结,与当今 LLM 的发展和我们取得的进步之间有着深层联系。
Lenny Rachitsky: 我记得是 Ilya 在 Dwarkesh 的播客里说过——你可能会觉得 LLM 不过是”预测下一个词而已,有什么了不起的”,但要做到这一点,它必须理解整个宇宙,理解这个世界上发生过的所有事情、存在过的所有事物、所有人写下的所有内容,才能预测下一个词。
Peter Deng: 对,说得太好了。
回到产品话题
Lenny Rachitsky: 好。那我稍微拉远一点,把话题转向产品本身。
Peter Deng: 当然。
反直觉的产品经验
Lenny Rachitsky: 你曾参与打造了一些历史上标志性的产品——你在 OpenAI、Facebook、Uber 工作过,还担任过 Instagram 的产品负责人。那我就直接问你这个问题,看看会引出什么。关于做产品或带团队,你学到的最反直觉、与常识相悖的教训是什么?
Peter Deng: 我觉得有一件事——这是我在 Uber 学到的一个非常艰难的教训——那就是有时候你的产品其实并不重要。我说的”产品”指的是你放在屏幕上的像素,或者你在移动应用里构建的那些东西。在 Uber,我之所以学到这一点,是因为——说来心痛——Uber 真正的产品其实就是价格和预计到达时间(ETA)。我觉得科技公司里的很多人往往把产品仅仅理解为那个数字化的呈现,但如果从整体视角来看,我们人类消费的是产品的全部。这是我学到的教训中最扎心的一点:有时候那些像素并没有你想象的那么重要。你修了一个 bug,我不是说你不该修,但它的效果远不如价格或 ETA 这样对用户更重要的东西来得大。
产品体验与商业逻辑
这种情况在 B2B 产品中很常见——不仅仅在于你的产品有多受终端用户喜爱,更在于它是否具备商业合理性。这是我在 Uber 学到的又一个硬道理,当时我还是一个满怀热忱、以设计为导向的产品经理。我前几天还有一个想法:当今最有价值的那些科技公司,说出来可能有些反直觉,它们中有许多在起步时并没有什么技术突破。它们确实建立在某种技术突破之上,后来也构建了大量技术。但这些公司中的很多——比如 Facebook——只是在早期阶段投入了大量苦功夫,本质上就是基于一个人际关系的数据库,在上面构建出了有价值的东西。
技术突破并非必需
然后不断打磨、迭代那个产品,推出新的功能,比如动态消息(newsfeed)和照片标注(photo tagging),这些其实就是来源于对用户需求的认真观察。其中有些想法极其简单,并不是从实验室里诞生的。再比如 Uber,它利用了每个人的口袋里都有一台 GPS 设备这一事实。他们没有发明 GPS 设备,但他们能够把这件事和”人们有车""人们需要出行""存在真实的人类需求”这些事实串联起来,把所有东西整合在一起。
最终,他们构建了大量技术来预测正确的市场供需和定价等等。但很大程度上,Uber 作为一家极具价值的科技公司,本质上是一家运营公司。我想向我在那里的同事们致敬——那些从运营角度负责 Uber Eats 和 Uber Rides 的同事。因为那真的是我见过的最重大的商业模式创新之一。所以在硅谷,这一点经常被忽略——人们会说,“哦,这是一家新的科技公司。“但实际上,一些最有价值的公司,恰恰就是在现有技术之上构建用户所需要的东西。
Lenny Rachitsky: 这里头有太多值得聊的了,我很喜欢。而且这番话出自一个领导过 Uber 乘客端产品团队、在 Facebook 工作过、还担任过 Instagram 产品负责人的人之口。从你这样的产品人口中说出,分量完全不同。
Peter Deng: 对,沿着 Instagram 的话题再说一下——它的想法极其简单:展示照片、视觉分享。但 Mike 和 Kevin 在打磨产品上投入的工艺和苦功夫,才是它真正腾飞的原因。这是一个很好的例子。我刚才居然忘了提 Instagram——怎么可能忘呢?它并不是其他公司做不到的事情,而是 Kevin 和 Mike 具备的那种产品品味和信念——他们确信人们想要某种特定的感觉,然后围绕这种感觉去构建和迭代。你看看现在,视觉分享已经成为我们生活的核心组成部分。他们真正解决了这个问题。
Lenny Rachitsky: 是的,我刚请 Mike Krieger 上过播客。所以我觉得这里存在两种张力。一方面,在很多非常成功的公司里,产品并不重要——它次于汽车、司机、GPS 和手机。另一方面,你不需要技术突破就能建立一个巨大的生意。几乎可以说,如果没有技术突破,那产品就变得重要了。Facebook 就是一个例子——本质上就是一个关系数据库。但让 Instagram 脱颖而出的,在当时有诸多竞争者的情况下,是体验——体验是不是好得多?反过来,如果体验不重要,那突破就体现在运营和其他方面。这个理解对吗?你说的基本上是这个意思?
Peter Deng: 是的,完全认同。我觉得两者都必须成立。但我也想说,即便你创立了一家拥有重大技术突破的公司,产品体验很快也会开始变得重要,因为技术优势能持续多久?用户迟早会回过味来——“这不是我想用的产品。我用的方式不太一样,这个更顺手”等等。所以我觉得你刚才的总结非常精辟。我也认为,在公司发展的不同阶段,什么更重要是会变化的。
AI 时代的护城河
Lenny Rachitsky: 这对于在 LLM 和 AI 基础设施之上构建的公司来说尤其有意思——你基本上在说,你不需要什么技术突破才能构建有价值的东西,只要你能够创造一种真正独特、出色的体验,来释放这种超级智能的潜力。
Peter Deng: 我觉得你说得对。关于那些在 LLM 之上构建的公司,我还有一些想法——这其实是另一个层面的问题。我认为对于它们来说,拥有正确的数据,以及正确的数据飞轮(data flywheel),至关重要。
Lenny Rachitsky: 比如专有数据。
Peter Deng: 没错。而飞轮的关键在于,你可以从专有数据起步,但飞轮的真正意义在于你如何持续维护和生成这些数据。第二点,又回到了工作流的问题——也就是它如何真正融入人们的生活中的使用体验。这一点会越来越重要。
Lenny Rachitsky: 那我们就多花点时间聊聊这个,因为很多人正在思考这件事。现在似乎人人都想创业,AI 大大拓宽了可能性。所以很多人很好奇,应该把精力花在哪里。我觉得这个话题真的很有意思。那我听到的是两个关键点——要建立任何形式的护城河,防御基础模型来抢你的饭碗、或者其他公司来竞争:一是你能获取什么样的专有数据,并建立飞轮来持续生成更多这类数据?二是你如何切入一个非常具体的垂直领域——你对它有深入理解,并且能够融入用户已有的工作流?对吗?我大概是说对了。
Peter Deng: 对,这个话题我们可以展开聊很久。因为无论你想做什么产品,总有现有玩家拥有分发优势。但我确实有一个观点——有些产品能够突破那些公司分发上的优势,但你的产品必须达到相当高的门槛,必须比对手好得多才行。我觉得这是第一点。
但数据飞轮这件事确实很有意思,因为模型会对你展示给它的任何数据变得非常擅长,这也是很多人把 AI 想象成一根魔杖的原因之一。但不是的——如果它在正确的数据上做了训练,它就会做好它被训练去做的事情。它的可塑性很强。
对于今天创业的人来说,非常审慎地思考你能获取什么样的数据来启动飞轮,以及你能做什么来让飞轮持续运转,这将是一个关键问题。
Lenny Rachitsky: 我们说得更具体一些吧。你说到这个的时候,我想到——Windsurf 的 CEO 来过播客,我们聊了很多他们拥有的那些非常独特的数据,比如人们接受和拒绝哪些代码片段推荐,他们实际上基于这些数据发布了自己的模型。这是一个好的例子吗?还有其他例子能说明这一点吗?
Peter Deng: 这是一个完美的例子。
我投资了一些尚未公开的公司,它们在这方面各有自己的做法,非常有趣——能够利用产品中的各种活动来让它们所做的事情变得更聪明。这也是为什么我认为数据飞轮和工作流如此密不可分,因为如果你正在为企业和人们解决真正有价值的问题,并且有大量的注意力被投入其中、大量的工作通过它来完成,你就会拥有这种优势。
我看到一些身处非常不同市场的创业公司拥有这种洞察力,对此有非常深入的理解,而不是试图零样本(zero shot)搞定一切,然后说”不不不,我们就是这样做的”——他们真正在思考的是如何把产品做得真正有用,这样它才能随着时间推移变得越来越有用。
这将非常了不起,因为作为这些产品的消费者,我们都会从中受益。
数据飞轮的构建
Lenny Rachitsky: 我从中还听到一点:如果你没有专有数据或独特数据,你仍然有机会通过构建飞轮来收集这些数据——通过用户的实际使用。
比如 Windsurf,如果他们一开始都构建在 Claude 3.5 之上,然后他们积累了所有这些独特数据,现在他们发布了自己的模型。
Peter Deng: 完全正确。
这回到了我之前可能简单提到过的一点——做任何事情都需要毅力(grit)。你必须拥有那种愿景,有明确的方向,并且能够真正去追逐它。我觉得这非常重要。
产品工艺与分发优势
Lenny Rachitsky: 关于你说的分发优势可以被克服的例子,我经常想到的一个好例子——我们请过微软的一位 CPO 上节目,结果发现微软有好多位 CPO,我之前都不知道他们有这么多——我问她,“为什么 Copilot 没有做到……” 世界上增长最快的公司——Cursor、Windsurf、Lovable、Bolt,所有这些。Copilot 远远走在了这些公司前面,但这些公司突破了出来。
虽然微软拥有分发优势、顶尖人才、基础设施,所有这些条件,以及先发优势——但正如你所说,Cursor、Windsurf、Lovable、Bolt 这些公司只是做出了更好的产品。
Peter Deng: 我确实相信存在一种产品工艺(product craft)的层级,能让用户觉得切换或尝试新东西是值得的。我看到一些产品具备这种特质。我觉得 Granola 就是其中之一。
Google Meet、Google、最初起步的 Facebook、Microsoft Teams、Zoom 都拥有如此多的分发优势,但 Granola 有这些微小的、令人愉悦的产品工艺细节,我个人非常欣赏——“是的,他们懂了。”
他们有这些小小的优势,恰到好处,他们真的找到了一种方式让它变得如此令人愉悦,以至于你会说,“对,我会安装这个软件。是的,百分之百我会跟朋友们推荐它,因为它太改变生活了。”
我们现在开始看到这种情况了。之前,大概 18 个月前,人们会说,“哦,谁拥有最好的模型?” 但现在,真正重要的是谁有最好的工作流、谁有最好的产品。而且我们人类是很挑剔的——我们想要最好的。所以当有人推出了精心打造的产品,人们会注意到的。
Lenny Rachitsky: 这里有几个要点:如果你想创建一家 AI 创业公司,有几件事需要考虑,这会让你更有可能突破并胜出——你的数据飞轮是什么,你如何收集专有的独特数据;你如何构建出产品工艺让人惊叹、愿意告诉朋友的产品。
Granola 就是一个很好的例子。显然 Cursor、Lovable、Bolt、Replit,这些公司都做到了。而且感觉他们对某个垂直领域的工作流有深入的理解,了解用户的问题,并以独特的方式加以解决。
Peter Deng: 对,我自己也说不了更好了。
产品在 AI 公司中的角色
Lenny Rachitsky: 太好了。我想问你一个问题,这在我和 Mike 在 Anthropic 聊天时提到过,也和这个话题相关。我在思考的是,Anthropic 的产品团队到底在做什么。
他们在构建一个基本上是超级大脑、超级智能的东西,它会无所不知,也许未来会自己构建体验。然后有一个产品团队在顶层构建一个交互层,来跟这个超级智能大脑对接。
Peter Deng: 那这一层的意义是什么?这一层的价值是什么?
Lenny Rachitsky: 你刚才说到了一些,关于体验的价值、让产品感觉原生的价值。但我还是想直接问你:在像 Anthropic、OpenAI 这样的公司,团队在做超级智能,上面有一个 UX 层——你认为产品方向会怎么走?
Peter Deng: 我认为这些公司拥有巨大的优势,因为你可以和研究人员在同一栋楼里工作。我认为在训练后(post training)和产品之间存在一种真正共生关系、紧密的伙伴关系。越来越多的,关键不再在于原始智能,而在于对模型能力的微调——让它真正与人们产生共鸣、满足人们的需求,以及产品的演进方向。我觉得你会看到这种趋势越来越明显。
我觉得这个不太关乎 Anthropic,更多是关于 OpenAI。我认为 OpenAI 做了一个很棒的举动。
我是 Fidji 的超级粉丝。她要加入的消息一泄露出来,我就给她发短信了。我说,“太好了。太棒了。恭喜。”
我为她、为这家公司、为我所有还在 OpenAI 的朋友们感到高兴,因为这样一位了不起的领导者要来了。
作为消费者我也很兴奋,因为会有一些优秀的产品问世。
我认为,在任何一个大型模型公司里,训练后(post training)团队与产品团队之间这种紧密无间的关系,都会催生出一些真正令人惊叹的成果。
Lenny Rachitsky: 首先,Mike 实际上说过非常类似的话,越——
Peter Deng: 我向你保证我没看过那期播客。
Lenny Rachitsky: 那期还没播出来呢,我相信你。
他有一个很有意思的发现:他把产品人员放在用户体验、面向前方的产品上,然后把项目经理放在研究团队中,帮助构建模型、帮助模型变得更好、帮助研究人员开发东西。他发现所有的杠杆效应和胜利都来自项目经理与研究人员的协作,在产品体验方面反而少得多。于是他把越来越多的项目经理配置到那个团队。
Peter Deng: 听到这个我太高兴了,因为那有点……非常有验证感,因为我们在 OpenAI 也是这么做的。我们与训练后团队紧密协作,正是因为那种紧密的合作,你才会看到 ChatGPT 在许多方面不断进步。太好了。我们独立得出了相同的结论,这太棒了。
Lenny Rachitsky: 是的。这是个好迹象。
从零到一再到规模化
好的,我们之前在谈论初创企业、创办新公司。我想沿着这条线再往下聊聊。
Peter Deng: 好的。
Lenny Rachitsky: 我觉得在你工作过的所有公司中,你从零到一再到规模化打造的产品,可能比大多数人都多。我来快速回顾一下你做过的一些事情,肯定会遗漏一些,但让我试试看。
你构建并领导了 Facebook 动态消息(newsfeed),也就是现在的版本。你打造了新的群组体验——聊天和消息功能。你把 Messenger 作为独立应用发布,那是你的项目之一。
你领导了 UberPool 低价出行项目。你推出了 ChatGPT Enterprise。你发布了语音和视觉功能、记忆功能、自定义 GPT,重新设计了 ChatGPT 的整体界面。还有很多其他的东西。
在 Airtable 显然也有大量工作。还有 Oculus。
这些只是引言中提到的一些例子。我接下来会逐一展开。
说了这么多,我觉得你在从创意到零到一再到规模化这条路上,见过太多什么有效、什么无效的经验了。所以我就直接问你这个问题:从创意到一再到数十亿用户,关于成功需要什么,你学到了什么重要的教训?
Peter Deng: 谢谢。你刚才念这些的时候,也让我好好回忆了一番。
我想说的第一件事是,从零到一和从一到一百是不同的。当你处于一到一百的阶段时——我大部分时间都花在这个阶段——我们在两年内把 Instagram 的使用量翻了四倍,那是一段非常刺激的旅程,在其他公司也有很多类似的例子。
但当你从一到一百的时候,我认为有一件事你必须认真考虑:你必须提前规划好你的棋步。你必须在行动之前深思熟虑,构建那些能让你可持续地加速的系统,因为从零到一阶段你是在寻找产品市场契合点,而到了一到一百阶段,你要确保自己能以最快速度达到超大规模。
我很有幸搭上了许多产品经历超大规模扩张的顺风车。我常喜欢用的比喻是,当你经历这个过程时,你能感受到 G 力。有些人会说,“哦,我是飞行员,我能在三万五千英尺高空飞行。“但感受火箭起飞时的 G 力是完全不同的。
我在这个过程中做过几次之后学到的一件事是:你必须构建那些帮助你可持续地加速的系统,有时候,你必须先慢下来才能跑得快。
举个例子。
在构建动态消息的时候——我们现在用的版本——从我们构建它到现在其实没有太大变化,大概十二年前左右吧,我不太清楚为什么没有太大变化。
但我愿意认为,那是因为我们投入了大量时间和工艺去思考整个分享循环:关键环节有哪些,架构是怎样的,信息架构是什么,整个流程是什么样的——从在页面顶部发布内容,到它出现在动态消息中,到有人点击”赞”,到通知图标亮起红色,然后这个循环一遍又一遍地重复。
我愿意认为,动态消息之所以经受住了时间的考验——现在这个版本——是因为我们非常仔细地思考了人们想要如何互动、想要如何消费信息,以及整个循环。当这一切都到位的时候,我认为产品就会经久耐用。我觉得这在很多不同的公司都是如此。
当我在 Uber 的时候,乘客端应用存在一种意大利面条式代码的状况,但退后一步重新架构——核心组件是什么,如何让产品选择器能在全球范围内扩展——这才是关键。
说一个鲜为人知的事情。说到毅力(grit)和苦干。
Uber 可不只是简单地找辆车那么容易。如果你去过别的国家,比如在印度,有时候根本没有路牌,所以你只能在这个小超市门口之类的地方上车。有整整一个团队专门负责上车点和下车点。这是一项大规模的工作。
听起来很无聊,但它对 Uber 的扩展能力至关重要,因为上下车团队思考的是,“那场馆该怎么办?“找到正确的抽象意味着你可以拥有一种可扩展的方式来处理机场的上车点,并配置不同的场馆。
这些系统——当你在一到一百阶段花时间去构建它们——会帮助你大幅提速,这就是你在两年内获得四倍用户增长的方式。
再比如 Messenger,我们在推送通知等基础设施上投入了大量思考。我们在大约两年半的时间里把这个产品从零做到了每天发送 47 亿条消息。我认为这确实需要那种前瞻性的思考来构建正确的系统。
Lenny Rachitsky: 让我顺着这条线追问一下,因为这真的很有意思。
你本质上是在说,一旦找到了产品市场契合点——其实在你开始规划之前我想先问你——当你开始从一到一百扩张的时候,你的建议基本上是不要”快速行动、打破常规”。不要发布 MVP。这是一个真正要提前想好很多步棋的阶段,想清楚你需要什么才能把这个产品做到——比如说——十亿用户。
Peter Deng: 对,对。是构建系统,然后那种系统性思维会带你走得很远,至少这是我的经验,希望你能找到同样的路径,不过具体情况可能因人而异。但没错,正是如此。
Lenny Rachitsky: 关于什么时候该这样做,你有什么指导建议吗?因为你构建了一个东西,好的,它奏效了,同时也会有这种想法,“好的,就让它继续跑下去,能扩展到多远就扩展到多远。“在你的经验中——关于什么时候真正应该退后一步,提前思考很多很多年,你有什么指导建议?
Peter Deng: 好问题。
首先我想说的是,这不是一个二元开关。它实际上是一个渐进的斜率。
投资组合思维与资源分配
Peter Deng: 在我带领团队的时候,我一直非常相信这种投资组合式的思路。众所周知,Google 有过著名的 70-20-10 投资组合方法。对于更成熟的公司来说,这可能是合适的;如果你是初创公司,也许是五五开。但你不能以二元的方式来思考这件事,而要以一种关于如何扩展以及何时需要投入更多资源的方式来思考。
每家初创公司都不一样。你发布的每款产品也不一样。然后去思考你的投资组合策略以及如何分配时间——这就是我的建议。这真的取决于你所在的阶段。
度量一切:建立增长团队的价值
我觉得这其实很好地过渡到我想说的第二点,如果可以的话——就是当你从 1 到 5 或者 1 到 10 的阶段——不是完全从 1 到 100——我发现有一件事非常有帮助,那就是度量一切。
这听起来又是一个非常简单的道理,但就像你不会在没有仪表盘的情况下驾驶飞机一样,为什么你要在不了解仪表数据的情况下运行你的产品呢?
在我带领过的几乎所有团队中——不管是 Instagram、Uber、Airtable——以及 ChatGPT,都是如此。我做的第一件事之一就是建立增长团队。
建立增长团队其实很有意思,因为它看起来是一个简单的准则,一个简单的事情。就是”我要建一个增长团队”,但接下来你会发现很多东西。
你会发现有多少东西你还没有打点日志,你在审视整个产品时有多么不严谨。
说来好笑,因为这部电影我看了太多遍了,同一部电影反复上演——我记得走进 Instagram 的时候,问 Kevin 和 Mike,“我们有多少用户?“回答是,“嗯,我们其实不太清楚。“就是说,“有很多,但我们真的不太清楚。”
当你建立增长团队并招到合适的增长负责人时——我有幸与 Instagram 的 George Lee、Facebook 的一些早期增长人员、Uber 的 Andrew Chen、Airtable 的 Lauryn(她现在在领导 Notion 的增长团队)合作过。我一直非常幸运,能与一些非常出色的人共事。
当你招到合适的人,他们就会问出所有正确的问题。因为增长产品经理这类人的典型反应会是,“等等,为什么会这样?让我们把 X、Y、Z 的数据拿出来看看。“这时候你才发现你根本没有记录 X、Y、Z 的日志。等你把 X、Y、Z 的日志打点好之后,你看着数据又说,“等等,这又是怎么回事?“于是你逼着自己更深入地分析,做一些相关性分析,提出假设。
因为增长负责人、增长产品负责人对实验这一面非常着迷,所以这其实是一件很容易做到的事——当你开始组建增长团队时,它会自然而然地催生所有正确的问题,然后逐渐转化为所有正确的行为:把你一直在做的、看起来在奏效的东西,变成一个更严谨的系统。
这就是——抱歉——从 1 到 10 的阶段,我觉得这真的为你后续的 10 到 100 打下了基础。
Lenny Rachitsky: 我喜欢这个增长团队建议的原因是,很多人想到雇佣增长团队的时机是”我们需要推动增长”。但你说的是,它有很多二阶收益——就是帮你搞清楚到底发生了什么,让很多人真正了解事情的运转状况。
Peter Deng: 完全同意。
我觉得我建议建增长团队而不是建分析团队的原因在于——因为如果你建一个分析团队或数据科学团队,很可能没人会听他们的。他们会说”我有这些洞察”,但”嗯,没人在乎”。
但如果你招一位增长负责人,他现在与推动增长的结果直接挂钩,所以他会是那个认真倾听、追问更多问题、并与数据科学团队深度协作的人,让你的整个产品和业务变得更加严谨。这改变了你整个团队的基因。
品味与工艺:增长与产品品质的平衡
Lenny Rachitsky: 我想聊聊招聘的事,但在这方面你还有什么想分享的吗?关于做新产品、扩展产品?
Peter Deng: 我想说的最后一件事是——我想提醒大家,有时候在追求数字的过程中,产品人会忽视品味和工艺的重要性。也许这恰好可以过渡到团队建设的话题——但你必须有对冲的力量。
非常重要的一点是,给团队中两个人不同的任务。一个人说”去增长产品”,另一个人说”等一下,保持那个设计、那个美感、你的产品赖以成名的工艺”。这种张力是非常健康的。我在 Facebook 见过,在 Instagram 见过,我在 Instagram 亲手创造过这种健康的张力。Airtable 也是一样——但关键是要有这种张力……ChatGPT 也是完全一样。
你必须同时在两边都有这种推拉,才能真正拉开全谱。
Lenny Rachitsky: 这就引出了一个问题——你到底怎么做到这一点?你可以嘴上说说,“好吧,我们要确保体验很棒,同时也要增长这个数字。这是你的目标。“你如何将其操作化?是放在绩效考核里?作为一项能力维度?还是靠文化,或者别的什么?
搭建”复仇者联盟”式的团队
Peter Deng: 作为领导者,你必须以正确的方式搭建团队。你必须真正把团队当作一个产品来思考——你需要哪些不同的部分,才能真正拉伸你思考的广度。
我参与搭建过的团队中……最成功的是那种”复仇者联盟”式的团队——每个人都非常不同,拥有截然不同的超能力,但作为领导者,你是那个帮助裁决分歧和不同意见的人,而你知道,当每个人各自痴迷于不同的事情时,你会得到最好的结果。这很重要。
创造你的平衡,真正拓展你审视的空间,创造那些健康的辩论——这很重要。
我觉得很多人忽视了这一点。有些人把团队里的人当成干活的”人头”,但我的理念一直是去思考:“公司需要什么才能成功?谁是在那一件事情上特别突出的人?我如何确保我们招到那个人?又如何确保我们招到另一个人,再另一个人?”
这几乎就像在玩 RPG 游戏,每个人都有不同的属性滑块,你必须打造这支超级队伍,让每个人在不同的方面都有突出的尖刺。
当你在那种环境中创造出那种氛围时,你的团队会走得很远——这是我获得过很多成功的经验。
Lenny Rachitsky: 这是一个非常有趣的回答,也是我没听到过的。本质上,不是创建正确的激励机制,而是招那些天然以特定方式看世界的人,这样就自然形成了平衡——比如产品经理、设计师和工程师之间的健康张力。
Lenny Rachitsky: 这真的很有意思,因为这种方式感觉比”这是你的目标,但同时也得确保体验够好、客服工单数量得降下来”要可持续得多。它就是自然而然地——这些人本身就想要这件事发生。
Peter Deng: 完全同意。其实我有一个框架……我认为有五种不同类型的产品经理,这个分类至今仍然成立。
这个框架最初是我在 Uber 时和同事们在一次随意的讨论中碰撞出来的。我们当时是为了优化招聘实践而梳理的。
我每到一家公司,都跟招聘人员是最好的朋友,因为说实话,我的核心理念就是必须组建正确的团队,所以我们必须深度合作。
在 Uber,我们总结了产品经理的五种原型。至今我仍然认为这个分类是准确的,到现在依然成立。这个有意思吧?你要我展开讲讲吗?
Lenny Rachitsky: 当然,我太想听听这五种了。
Peter Deng: 这五种是我发现最经得起时间检验、彼此之间也最有区分度的类型。
产品经理的五种原型
你之前说得很到位,Lenny——当你招对了人,他们天然会被不同的东西驱动。我们总结出来的就是这五种。
第一种是消费者产品经理(consumer PM)。这种人一半是设计师,一半是产品人,对细节极其执着。“体验够不够愉悦?工艺够不够精?天哪,这差了三个像素,我受不了了,快把我逼疯了。为什么这么复杂?“这些人有时候会被认为是那种吹毛求疵型的 PM,但这只是其中一种。
另一种就是之前提到的增长产品经理(growth PM)。这种人一半是数据科学家,一半是产品人,他们天生就先想数字。最优秀的增长产品经理有一种气质:“我很持怀疑态度,给我看数据,我们跑个测试来验证。我不信你。“我在框架里先说这两种,因为它们确实截然不同。一个是”我有感觉,我感觉到了,这个更好”,另一个是”不,我不信你,我们应该测试验证一下。“这种张力非常健康。
我最喜欢把这两个人放在一个房间里辩论。我会说,“太好了,我们一定能做出好东西,把这个产品推向前进。”
第三种是我称之为 GM PM 或商业产品经理(business PM)的类型。这种人一半是 MBA,一半是产品人。他们天生就从商业模式出发去思考——“利润率是多少?机会在哪里?价值在哪里被创造?”
在 Uber 我们有很多这种类型,他们是市场产品经理,就是……
我很喜欢跟他们合作,因为他们的思维方式完全不同。他们思考问题的方式是,“嗯,这里的激励机制是什么?“这是一种非常迷人的思维方式。
更多 PM 原型
还有一种,实际上比你想的更微妙,我称之为平台产品经理(platform PM)。这种人天生就致力于为其他人构建工具。
在 Uber,我们有用于消息推送的内部平台,或者用来构建内部工具的平台。
这些人经常被忽视,但这其实是一种非常深层的内在驱动——因为他们要构建的是让你跑得更快的系统,而这正是他们热爱做的事。
最后一种,我以前叫算法产品经理(algorithms PM),但在 AI 时代,我会把它改名为研究产品经理(research PM)。这种人一半是研究员,一半是工程师,一半是产品人。这些人的头脑非常惊人。
基本上,他们就是传统意义上 Google 搜索算法产品经理那样的角色,但在当今,他们就是那些既有产品品味,又深入理解技术和模型训练方式,从而能够影响技术方向、构建出最出色产品的人。
就是这五种。
至今我仍然认为这个分类是成立的。也许我们那天在 Uber 头脑风暴的时候确实发现了点什么。不过我也很想听听你的反馈。
Lenny Rachitsky: 这太好了。听你说的时候,我脑子里就在想,“这个人对应那个,那个人对应这个,哦,他们归这里。“非常有共鸣。
(广告段落已跳过)
简单总结一下,就是消费者产品经理、增长产品经理、商业/GM 产品经理、平台产品经理和研究产品经理。
Peter Deng: 对。
Lenny Rachitsky: 很多人现在管最后那种叫 AI 产品经理(AI PM)。我觉得这个叫法现在已经很流行了。
Peter Deng: 你得与时俱进。对。
不过这个框架里还有一点我觉得很有意思——每个人都有一个主导类型和一个辅助类型。
它就像那些人格测试一样。也许我们当初这么做只是因为很难把人归类到单一框里,我自己也不觉得能被简单归类,但我确实认为人们以一种思维方式为主导,同时也有一个辅助的方面来保持平衡。
如果你认可这一点并把它应用到团队中,我很想听听你的听众们是否也有共鸣。也许这个框架能帮你们意识到——你可能缺了一个不该缺的人。
Lenny Rachitsky: 你自己做产品经理时的原型是哪种?
Peter Deng: 这也是人格测试的另一个特点——你听到某种类型时会说,“这就是我,我认领了。”
毫无疑问,我是消费者产品经理,同时也是增长产品经理。我主要偏消费者……我没办法……
我之前跟你聊过我喜欢的那些产品。我能看到人们投入的细节,我由衷地欣赏。但归根结底,“我们得去衡量一些东西。“这就是我。不过话说回来,每个人都不一样。
Lenny Rachitsky: 我很喜欢你说的这点——很多人想到产品经理,听到第一种例子就觉得,“哦,我大概得是那种类型,因为大家谈论优秀产品经理时说的都是这种。“但你的意思是,成功的 PM 有很多种路径。
我在 Airbnb 的时候做过一次人格测试,最大的收获之一是——它是一个颜色测试,你会得到绿色、黄色、红色之类的,而我们团队在光谱上是完全分散的。这很好地提醒了你,你可以是不同类型的人,在产品经理这个角色上依然可以非常成功。
这大概就是因为有不同的原型存在,而 PM 的需求和角色也各不相同。“产品经理”这个名称之下,PM 们做的事情其实有很多种。
创始人与市场的匹配
Peter Deng: 另外,作为投资者,现在我觉得很重要的一点是看创始人与市场的匹配度。因为如果你把一个面向消费者的 PM 放进一个非常枯燥的受监管行业,他们很可能会感到沮丧,很可能坚持不下去。而有些人,你看到他们的路演就会觉得,“哇,你真的对这个问题充满热情——”
你看到路演就觉得,“哇,你真的对这个问题充满热情,你真的在乎为他人打造工具,这正是……”这就是 Twilio 式的 PM,或者其他什么类型。“你和这个业务完美匹配,太棒了。“所以我觉得,我很喜欢你刚才总结中说的那番话,因为我认为做 PM 没有唯一正确的方式,希望这个框架能给大家多一点空间,去展现真实的自己。
Lenny Rachitsky: 我很好奇其他职能是不是也有这样的原型分类,比如设计师、工程师,不过这个就不展开了。如果你在 YouTube 上听这期节目,留个评论说说你觉得自己是哪种原型吧。你的主要原型和次要原型分别是什么?我再念一遍:消费型 PM、增长产品经理(growth PM)、商业/总经理型 PM、平台型 PM、研究/AI 型 PM。
Peter Deng: 太好了。
招聘之道
Lenny Rachitsky: 好,我想聊聊招聘。这个话题其实在我和你之前的同事聊天时经常出现,特别是 Nick Turley,他是 ChatGPT 的产品负责人,我们正在努力邀请他来上播客。因为——
Peter Deng: 是的。
Lenny Rachitsky: 那可是——
Peter Deng: 他很棒。
Lenny Rachitsky: 我也听说了。他告诉我,ChatGPT 现任的工程负责人、首席产品工程师、设计负责人和市场负责人,都是你招进来的人。而且你招的很多人后来都做出了非常了不起的成就。你已经分享了其中一些人的名字,他们中有不少上过我的播客——这可是成功的终极衡量标准。那我就直接问了:在招聘时,你关注的某一个东西是什么——那种别人没足够重视、被低估的特质——能帮你发现那些出色的明星人才?
Peter Deng: 听到 Nick 这么说真的很荣幸。Nick 是我共事过的最优秀的人之一,毫无疑问。事实上,我想借此机会说一句,OpenAI 的人基本上是我职业生涯中合作过的最优秀的一群人。当初接受这份工作的时候,我告诉团队:“这会是我最后一个运营角色,我会倾尽全力,全力以赴。“基本上,我在招聘和搭建团队上花的时间,大概和思考产品的时间一样多,甚至更多。这也呼应了我之前说的——你必须把合适的人聚在一起,才能产生巨大的影响力。而很多时候领导者会忽视这一点,觉得”随便找个人就行”,但实际上,在某些领域有专长的人,能和在其他领域有专长的人形成互补。当你把这样的团队搭建起来,奇迹就会发生。这是你能做的最好的投资,回报会非常丰厚。
所以我想这算是我的开场白——你必须……所有正在听节目的人,你得审视团队中的每一个人,看清楚你需要什么,然后在每个位置上找到最合适的人。说实话,在 OpenAI 的告别晚宴上,我最后说的大概就是:“听着,我都不知道这之后我还能做什么,因为我合作过的最优秀的人都在这里。“我们有 Ian Silber 负责设计,Thomas Dimson、Joey Flynn、Ryan O’Rourke。Nick Turley 是我在那里认识的非常出色的人。Joanne——我遗漏了太多人——还有负责产品营销的 Coley,负责营销传播的 Antonow,名单还可以继续。产品运营团队也非常出色。说实话,我对自己在那里搭建的团队感到骄傲,甚至超过了对产品本身的骄傲。所以我想说的是,这是我非常看重的一件事,也希望更多领导者能重视起来——真正用心地搭建你的团队,把它当作一个产品来对待。你必须精心打磨,你必须真正在乎这个团队。
Lenny Rachitsky: 在你说下一个建议之前,我想再强调一下这个观点。我真的很喜欢这个回答。如果我问别人,“你招聘的秘诀是什么?你会关注哪些别人不够重视的特质?“大多数回答都会聚焦在那个人身上——你应该关注什么特质,应该问什么面试问题。但你到目前为止的宏观回答其实是,关键不在于某个人,而在于团队会是什么样子,我们需要在哪里有突出的长板?我们需要如何平衡这支”复仇者联盟”的阵容?
Peter Deng: 完全正确,完全正确。话说到这里,我想,算是符合我的风格吧,我有两点关于招聘正确团队的想法想分享。我有这么一句话,实际上我有一份文档,叫”PXD API”,我带到过好几家公司,内容大概就是”以下是如何与我合作”。里面有一句话,是我在支持每个人和招聘每个人时真正优化的准则,那就是:六个月后,如果我还在告诉你该做什么,那我招错了人。这句话在三个层面上对我非常有用。第一,这是对我自己的提醒——无论是招聘还是寻找人才的时候,要保持极高的标准,绝不能将就。因为如果我降低了标准,大概率六个月后,我就没办法让这个人独立运转,还是会手把手告诉他该做什么——这不是我想要的。这不是我的初衷。
第二个作用是,我会在新人加入团队或者我们确定招聘的时候对他们说这句话,因为这向他们传达了我的标准,让他们知道怎样才算成功,也给他们一个努力的方向。
第三点是我们双方共同的事情——它帮助我和对方在一个更高的层面上协作。目标不再是”你有没有完成这个 OKR,有没有达成这个目标”,而是更高一层的目标变成:我们的校准够不够?我们是否正在走向这样一个状态——六个月后,是你来告诉我需要做什么?我们在往那个方向走吗?
因为如果以此为框架,那么双方犯的每一个错误,都变成了一次学习的机会——我们如何从中成长,达到六个月后我们想要的状态?我作为管理者,怎样做正确的事来为这个人创造成功的条件,使得六个月后我不再需要介入?
我认为这三点,以及”六个月后,如果我还在告诉你该做什么,那我招错了人”这条简单法则所带来的二阶效应——它给我施加了压力,也给对方施加了压力,同时创造了一个非常有趣的环境,一个可以安心思考”我们是否在朝着那个目标前进”的安全空间。再强调一次,在我待过的每一个地方,尽管我热爱打造产品,但我在搭建团队上同样感到无比自豪,这一直是一件非常快乐的事情。而这就是我在招聘方面的两个秘诀之一。
Lenny Rachitsky: 这太棒了。我有一个追问,但我想先指出为什么我觉得这个方法如此精妙——它的背后有一个隐含的前提,就是你可以信任这个人。所以它实际上是在问:我信任这个人吗?我觉得他会主动吗?我觉得他会有正确的判断力,本质上就是品味和直觉?这些才是这个问题之下更底层的考量,这一点非常好。而且这种自主性,感觉自主性几乎隐含了你想要招聘的人身上所有重要的特质。我喜欢这个问题对你和对方来说都如此简洁——
Peter Deng: 谢谢。你关于自主性的那番话,我非常认同。因为作为领导者、管理者,你的目标是实现规模化。如果这个简单的说法不成立,你怎么能打造出最优秀的公司、最优秀的产品?
Lenny Rachitsky: 那我的追问来了。这个方法主要适用于领导者吗?比如 ChatGPT 的产品负责人之类的,假设一个人不是 CPO,只是一个 PM 团队的经理,你觉得这个方法对他们也有用吗,还是说主要还是给领导者用的?
Peter Deng: 我认为这适用于所有人。适用于每一位管理者。因为如果你想在任何公司成为一名成功的管理者或领导者,不管你是从一线经理做起还是其他什么岗位,如果你想成长,或者哪怕只是——如果你在公司早期加入,你拥有大量的组织知识。那么在传递你所积累的经验智慧方面获得更多杠杆,对于取得成功至关重要。所以我认为每一位管理者都应该以这种方式对待自己的下属。因为它确实对所有人都有好处。对公司来说,拥有更多的杠杆和规模化能力是好事。对加入团队的人来说也是好事,因为他们知道成功是什么样的,并获得了一条持续成长的道路。对你作为领导者、管理者来说同样好,因为你能够实质性地提升整个团队的专业能力。
Lenny Rachitsky: 我觉得你甚至不需要真的打算不去告诉他们该做什么。它只是一个很好的视角,用来判断:这个人会不会很出色?即便你实际上打算某种程度上还是告诉他们该做什么。
Peter Deng: 对,没错。另外一点是,在你的面试过程中,你自然会去寻找这些洞察,去观察那些行为表现——他们是否真的有可能在六个月后做到这一点?这会在选拔环节给你一个非常好的视角,而不仅仅是在培养环节。
第二个招聘秘诀:成长型思维
Lenny Rachitsky: Peter,你的第二个秘诀是什么?这是一比一的配对。
Peter Deng: 好的。第二个我想说的是,我对此感受非常强烈——我最看重的品质是成长型思维(growth mindset)。这个认识是在我在 Facebook 的管理生涯中某个时刻形成的,当时我确实犯了一个错误,招了一个缺乏这种成长型思维的人。那真的非常困难,因为我的说法是:“听着,我没有时间对任何反馈进行粉饰。“坦率地说,我合作过的最优秀的人,是那些在一对一会议上对我直言不讳、告诉我哪里搞砸了的人。我喜欢这样,因为没有任何话是没有说出口的,我们能够推动事情向前发展——“嘿,我们怎么从中变得更好?“我觉得成长型思维就是这样一种东西,Lenny,在某个年龄之后真的很难再去教导。这对我来说以及我的家庭都非常重要——我对自己、对我的孩子、对工作中的同事都期望拥有成长型思维。
因为它创造了一个环境,每个人都乐于思考:我能进步的那一件事是什么?“每天进步 1%“这句话就真的能变成现实。有趣的是,每当我加入像 ChatGPT 或 Uber 这样的团队,我总是做我团队中候选人的最终面试。我会与招聘团队合作制定评分标准,我总是坚持做最后一轮面试。而我做的不是产品感,不做设计,不做执行,不做指标,我只做成长型思维。
有人会说,这不是很疯狂吗?其他所有这些能力呢?我的回答是:“我相当确定我可以信任其他人来评估那些能力。“但我觉得成长型思维对我来说如此重要——我们要建立一个团队,让每个人都自我反思,渴望进步,接受反馈,也给予反馈。这就是我发现的一个元层面的解锁。说真的,如果你没有成长型思维,不对反馈持开放态度,不乐于学习,那这就是一个元层面的阻碍。到那个时候,接受反馈会变得很困难,学习新技能会很困难,以任何有意义的方式发展自己都会很困难。所以我觉得这是真正关键的一环。
Lenny Rachitsky: 你刚才说的这件事分量很重。基本上作为一家公司的 CPO、产品负责人、高级产品领导者,你的面试不是”你是不是一个出色的产品经理?你有没有产品品味?“这类问题,而是成长型思维。
Peter Deng: 我想澄清一下,这是因为其他所有方面都已经由设计师、工程负责人等人面试过了。之前那个原则在这里也发挥了作用——我确实信任我的团队去评估那些人,但我最在乎的就是成长型思维。这就是关键。说实话,我确实也会做一些扫尾检查。如果我们在某个领域收到了弱信号,我会去考察一下。但我最后一轮面试的核心焦点一定是成长型思维。
如何识别成长型思维
Lenny Rachitsky: 好吧,那我得问问你在面试中具体怎么考察。不过在此之前,你提到成长型思维的时候,我脑海中浮现出 Marc Benioff 上播客时的画面。我问他,现在一切都在不断变化,在这样的世界里领导一家公司太疯狂了——每个人都在颠覆彼此,AI 在改变一切,速度太快了,每天都有新突破,你得不断跟进——你怎么应对?他说:“你应该想,‘很好。这太棒了。这是最好的建设时代。机会这么多,太令人兴奋了。这正是我们想要的。’”
Peter Deng: 没错。
Lenny Rachitsky: “很好。“我只记得他说了,“很好。”
Peter Deng: 我太喜欢这个了。
Lenny Rachitsky: 我觉得那就是成长型思维的缩影。
Peter Deng: 完全同意。
Lenny Rachitsky: 好,那我直接问你——你是如何辨别一个人是否具有强烈的成长型思维的?有什么方法?
Peter Deng: 好在我现在不再是运营者了,因为我要把我的面试问题公开了,这样就没人能作弊了。我觉得这也是现在做这期播客的另一个好理由。我问的问题多年来一直是同一个。你真的可以从这个问题中把成长型思维辨别出来。我请他们想一个自己犯过的最大的错误——真的,越痛苦越好。然后告诉我这个错误是什么。描述一下当时的情境,告诉我你现在实际上有了什么不同的思考方式、不同的工作方式。这如何变成了你的一个核心原则,等等。
我会给他们一点时间思考。有时候如果需要的话,我甚至会分享一些我自己犯过的错误。这很有趣,因为这个问题我问了太多次了,如果他们不是真心实意的,我能闻出其中的水分。
Peter Deng: 那种回答就像是,“哦,我工作太拼了”,或者”我做了某件事”。他们其实并没有真正……你能感受到人们愿意展露的脆弱程度。我也会对等回应,如果他们问我犯过什么错,我会如实告诉他们。这就是那种氛围。
但实际发生的是,这个问题之所以非常有意思,有多个原因。第一,你能感受到他们有多强的反思能力。有一次我和一位女性候选人聊,我们实际上聊了一个小时,因为她就是在给我讲她犯过的一个很精彩的错误,以及这个错误如何改变了她的工作方式,也改变了公司的工作方式。真的很了不起。你能感受到那份热情,你能感受到什么是真诚。当然也总会有一些人比较防备,不太愿意敞开心扉。环境是安全的,是一对一的,所以这是一个安全的空间。而且我认为这实际上不会对内向者或外向者产生偏向。我觉得到了那个时刻,一切都是非常真实的。第二个次级效应是,如果他们最终加入了团队,你们已经有过那个时刻了。你们已经有过那个时刻,已经说过了,“嘿,这是我真的搞砸的地方。“然后呢?一切都没问题。这不是损失,而是教训。这样就为你未来的工作关系设定了一个不同的基调。当然,我从来没有做过 A/B 测试,所以我没法告诉你这是否真的有效,但在我自己的工作风格中,我发现能够在那个层面上建立连接是非常有帮助的,无论是和直属下属还是和纽约的某位同事。
Lenny Rachitsky: 我喜欢这个回答的地方在于,它和我们这个播客的”失败角”非常像,那是我们播客的一个常设环节。我可能会调整一下”失败角”的形式,让它更接近这个问题。好,让我总结一下你找到的那些年来帮你发现明星员工的两个核心问题。第一个是,你会在六个月后告诉人们,“如果还在由我来告诉你该做什么,那我招错人了。“或者,你当面跟别人说的时候是怎么表述的?是不是直接说,“你可能不是适合这个岗位的人”?
Peter Deng: 实际上措辞略有不同。我的 API 中有五个部分,或者说是如何与我最佳合作的指南。与我共事最成功、我最喜欢合作的人有五个特质。其中一个就是表述为:是你在告诉我该做什么,而不是反过来。
Lenny Rachitsky: 入职六个月后。
Peter Deng: 对,对。然后我会跟进一句,“六个月后,如果还在由我来告诉你该做什么,那我招错了人。”
Lenny Rachitsky: 明白了。
Peter Deng: 我觉得,大概就是这么表述的。
Lenny Rachitsky: 顺便说一句,你应该把这份 PXD API 文档开源出来。
Peter Deng: 我很乐意。我觉得现在我已经没什么可藏的了。我就是,“来吧,我完全敞开的。“也许我们某个时候会这样做。你会让我有勇气这么做的,也许在这期播客播出之后。
Lenny Rachitsky: 那大家可能会在这期播客的节目说明里找到这个文档的链接。
Peter Deng: 如果我有勇气的话。
Lenny Rachitsky: 好。然后你问的另一个问题是,“给我讲一个你失败的故事,一个你发布的产品失败了,以及这如何改变了你的行为方式、你对产品的思考方式、你的运营方式。”
Peter Deng: 对。
Lenny Rachitsky: 非常好。好,我们来聊聊管理吧。
Peter Deng: 好的。
管理能力与向上管理
Lenny Rachitsky: 这个话题是这样来的——我和一些与你共事过的人聊过,有趣的是,出现频率最高的主题之一不是关于 AI 的……招聘也提到了一些,但最多的其实是关于你作为管理者的出色能力。这一点已经在我们之前聊过的很多内容中体现出来了。所以我想谈谈这里的几个方面。
Peter Deng: 好的。
Lenny Rachitsky: 一个是你曾经在 OpenAI 共事过的人,Joanna Jang?还是 Yang——
Peter Deng: Joanne?Joanne。
Lenny Rachitsky: Joanne。Joanne Jang,还是 Yang?
Peter Deng: 是 Jang。
Lenny Rachitsky: Jang。好的。你在 OpenAI 和她共事过,她分享了几件我觉得很有意思的事情。其中之一是,你教会了她如何更有效地向上管理,对她的职业生涯产生了深远的影响。你教了她一句很简单的话,她到现在一直在说、在用。首先,你还记得那句话是什么吗?
Peter Deng: 我说过太多话了,已经有些忘了。我常常忘掉自己说过什么,所以你可能得提醒我一下。
Lenny Rachitsky: 好,她说那句话是”承诺你会去做那件事,去做那件事,然后说你做了那件事”,作为向上管理的技巧。聊聊这个吧,这句话的力量和背后的含义。
Peter Deng: 其实,这是我在 Uber 期间学到的,来自负责公关、传播和政策的 Jill,她以前有句口头禅,叫作”重复不会破坏祷告”。道理很自然——大家都很忙。所以无论你考虑的是向上管理,还是管理整个团队,如果你不重复你的目标是什么,不重复你的愿景是什么,不重复你认为自己在做的那些重要的事情——无论是向你的经理汇报,还是别的——第一,你可能会忘记真正重要的事情。我认为这有点关乎行为习惯。这又是一个语言影响思维的例子。通过把这句话给 Joanne,也许就是想说,“嘿,让我们对自己要构建的东西更加有意识。“这就变成了一个持续的提醒。
这还有一个效果——如果你在说”这是我在做的事情”,然后你的经理可能会说,“等等,我们不需要再做这个了”,你就可以就此展开对话。而不是默默地做了却不告诉任何人你在做这件事。
让我退一步来说。第一,先说你打算做什么。然后在这个过程中,你就能够与你的经理——其实和任何人——对齐:我们到底要做什么?我认为措辞在这里非常重要,回到我之前说的,弄清楚那个目标是什么,并且精心打磨它,让它包含最大的信息量和最密集的概念。然后告诉他们你在做了,这是第二个阶段,在一对一沟通或团队全员会议上,你说”这是我们在做的事情”。这是一个很好的时机来确认你在推进,或者引出”这已经不是我们该做的事了”的对话。
最后你得告诉他们你做完了。把闭环合上,就是说,“好的,完成了。“我觉得这又是一个简短精炼却包含众多行为层面次级效应的短语。这有点像帮助他人的一个小技巧。有趣的是 Joanne 把它理解为向上管理,确实如此,但在我心目中,这更像是我们的运营方式,是我们如何成功保持专注、保持目标导向,并且能在目标不再相关时重新审视的方法。
Lenny Rachitsky: 那句话再重复一下,就是承诺你会去做那件事,去做那件事,然后说你做了那件事。
Peter Deng: 抱歉,再来一次。我自己的说法是:说你要去做那件事,说你在做那件事,然后说你做完了。
Lenny Rachitsky: 这也适用于做演示的建议。之前有人提过类似的说法,好像是谁说的来着,就是关于如何做好演示,也就是:告诉他们你要说什么,说出来,然后再告诉他们你刚刚说了什么。
Peter Deng: 我有可能确实是从那里吸收来的。所以这句话的原创权不归我。我只想说,是的,我确实是在重复这句话。
Lenny Rachitsky: 这个特别好。而且我很喜欢这不只是向上管理的建议,而是对所有人都适用的做事方法。最后一层含义是,确保别人知道你做了什么,某种程度上就是确保你获得应有的认可,让人们理解你创造的价值。
Peter Deng: 这确实很重要。我觉得有很多人性格比较内向,不愿意引人注目,没有那种英雄情结。这些人往往会在组织中被埋没。所以如果你是这个类型,记住要把自己做的事说出来。
Lenny Rachitsky: Joanne 还分享了你的另一个管理特质,我想多聊聊这个。就是说你特别擅长帮助人们认识到可以发挥自己的优势,而不需要把自己硬塞进某个固定的框架里。她说你基本上帮她在 OpenAI 内部创造了一个之前根本不存在的全新角色。也许可以分享一下这个例子,然后谈谈为什么这很重要,你是怎么想的。
Peter Deng: 我很高兴我们在聊 Joanne 跟你说的这些,因为 Joanne 真的很特别。我想借这个机会好好表扬她一下。她是我合作过的唯一一个,技术深度和产品品味(product taste)同样出色的人。我想在这里停顿一下。这真的很特别。我感到非常幸运,能在 OpenAI 与她共事。我从她身上学到了很多。说到六个月后才告诉你该做什么——她从第二天就开始告诉我该做什么了,而我非常喜欢这一点,因为她技术能力那么强,又有这样的品味,这两样东西结合在同一个人身上非常罕见。而 Joanne 正是因为在这方面的独特性,我注意到了这一点,就觉得,“哇,我合作过那么多产品经理,这个人真的很不一样。”
感觉我们必须想办法把这个角色塑造出来。于是我跟她说,“嘿,你能不能把这个东西写一份职位描述出来?因为这里面有一些神奇的东西,但我还不能完全理解它。“我觉得没有其他人会这样思考问题,而这可能成为 OpenAI 的一大优势。让我们把它固化下来。“再说回我一直强调的——语言是非常重要的——我觉得有时候,把你直觉感受到的东西写下来,就能得到一个可以传达给别人的载体。在这个案例中,Joanne 把她真正兴奋的事情写下来,帮助我真正理解了这些东西。而我恰好处于一个可以拍板的位置,于是我说,“看,让我们创建这个角色吧。让我们创建这个角色,由你来领导。我觉得如果我们能把它固化下来,对产品会非常好。”
所以我觉得我没做什么特别的事。我只是在跟随直觉,跟随她的引导。我要再澄清一下,那份文档不是我写的。我的记忆是,是她写的。所以所有这些事情中辛苦的工作都是她做的,我不想抢任何功劳。我唯一做的就是轻轻推了她一下,“我觉得这里有东西。你能不能花点时间把它写下来?“当她写出来之后,就是那种感觉,“好,这必须成为一个角色,而且你必须成为这个职能的负责人。”
Lenny Rachitsky: 她最终担任的角色具体是什么?我觉得分享这个会很有意思。
Peter Deng: 那个角色叫模型设计师(model designer),她对这个角色的定义方式非常有意思。我知道这个角色在其他基础模型公司可能以某种形式存在过,但她描述它的方式,以及她发现的所需的核心能力,让我们在招聘后招到了第一批两位模型设计师,他们和团队完美契合。而我认为,至少在我偏爱的视角下,这是我喜欢 ChatGPT 的一个重要原因——模型呈现出来的感觉、模型的气质,很大程度上归功于她创建并领导的这种技术加品味的角色。
Lenny Rachitsky: 我很喜欢这里的一个启示:作为领导者,就是要关注人们真正热爱什么,然后迈出那一步——让他们试着在文档里把想法清晰地描述出来。回到你说的语言和文字的力量,就是:“好,把你想的完完整整告诉我,我们一起打磨,也许这里面真有东西。”
Peter Deng: 对。
Lenny Rachitsky: 关于发挥优势这件事,你有没有更宏观的思考?因为有很多人在争论:我应该去补自己不擅长的东西来提升自己,还是找到自己特别擅长的东西然后在这方面做到更好?你怎么看?
匹配是双向的
Peter Deng: 我真心认为匹配是双向的。你所热爱的、你的优势所在,你需要找到与之匹配的公司和角色。我觉得很多人想要把自己硬塞进某个固定的模板里。很高兴我们聊到了产品经理的类型划分,希望这能让大家更自由地去投入自己真正热爱的事情。因为人生挺短的。如果每个人都能找到自己真正想做的事,然后全力投入去做,那就太好了。而我之所以对我们所处的时代感到乐观,也是因为现在有那么多不同的公司在涌现,所以总有能引起你共鸣的东西。
你想想我们正在做的事情就知道了——播客在二十年前根本不存在。但现在,我们拥有这些惊人的工具和平台,让人们能够真正表达自己,去做真正让他们快乐、同时也为世界创造巨大价值的事情。所以我认为,是的,我确实相信要发挥优势。我也认为,虽然有时候很困难,你需要审视自己现在所处的位置,问自己:这真的是你想做的事吗?还是有什么其他东西在吸引你的注意、牵引着你?
管理经验
Lenny Rachitsky: 还有一个管理相关的问题想问你。这个问题来自 Eric Antonell,他似乎跟你共事过十七年,跨越了很多不同的——
Peter Deng: 对,断断续续十七年。他是我最重要的导师之一,也是好朋友,非常厉害。
Lenny Rachitsky: 好。所以他说,“你一定要问这个问题。“他的原话是,你招聘、管理、指导过非常非常多的产品人员,有初级的也有资深的,跨越了很多不同的文化,他就说,“我们需要从你的经验中学到点什么”,关于你从中学到的——怎样才能成为一个真正成功的产品人,无论是在做产品方面还是职业发展方面,你从看到这么多不同类型的人、不同文化、不同资历中得到的那个核心心得是什么。
Peter Deng: 我觉得作为一个产品人,特别重要的一点是要对工艺细节极度痴迷。因为归根结底,你是在打造一个产品。要对工艺细节极度痴迷,同时又要具备一种视角和智慧——知道哪些细节其实并不重要。我想在这里停一下,试着——
我想停一下,试着把这点展开说一说。因为做一个产品人的核心,就是——你会想,哦,我要做一个大家真正喜欢的东西,这就是这份工作,这也是吸引人们来做产品人的原因——你内心有这种创造的欲望。我觉得我参与过足够多的团队,包括我自己,在我很年轻、刚开始做产品人的时候,我会对那些小细节非常痴迷,后来才意识到我们在一些其实无关紧要的事情上浪费了大量时间。所以我觉得这种二元对立对我来说很有趣,也很美妙,因为它概括了一个成功产品人的精神内核——你真的必须在乎,你必须对你正在打造的产品上心,但同时你又必须有足够的视角和商业判断力,知道应该把时间和精力花在哪里、把这份在乎用在哪里。
我自己感觉我经历了很多轮这样的循环。我做的每一件事,都会先极度深入、极度痴迷,然后退后一步看,发现等等,其实我忽略了什么,另一件事才更重要。我给你举个例子。用我在 Uber 的经历来说,我之前说过数字产品没那么重要,关键的是价格、预计到达时间(ETA)之类——我在 Uber 打造的产品之一是 Uber Reserve。它是最简单的东西。回到我之前说的,有时候最好的产品就是最简单的东西。但我们试图解决的问题是这样的:每个人都有这种经历——你有一趟早上六点的航班,你真的打算凌晨四点起床叫一辆 Uber,然后祈祷那时候有足够多的车、司机会来吗?
因为如果你这样做,你根本睡不好,每隔两个小时就会醒来,而且你可能还是会误机,因为你会睡过头什么的。所以就有了这样一个洞察:人们真正想要的是安心——知道车一定会来,这和现实之间存在巨大的错配。而且你知道吗?我愿意为此付费。于是我们打造了 Uber Reserve,它是最简单的东西——你只要告诉我们你的航班时间,我们会倒推计算,或者你直接告诉我们想什么时候被接走。这个产品的方方面面,我们都围绕用户真正需要的东西来打磨,那就是安心。所以如果你去那里,输入你的航班时间和接载时间,我觉得这个产品——自我离开以来,它没有太大变化。
它会告诉你,哦,这个时间非常紧,你可能赶不上航班。就像,哇。同样,这个功能就是基于”安心”这个原则放进去的。另一边呢,司机需要什么?他们需要知道你不会取消,还有其他各种顾虑。所以你也得考虑司机的激励。这是一个简单的想法,我非常为团队感到骄傲——他们搞清楚了所有复杂的细节,做了测试,我上次听内部的人说,这已经是一个年营收 50 亿美元的业务了,而且是利润率最高的业务之一。我真的很为此骄傲,因为它源自一个理念——让我们专注在真正重要的事情上,那就是安心,以及在那一刻有多少人真正需要它。我觉得这是我能讲的最好的故事了。
Lenny Rachitsky: 这个故事太棒了。它把你之前谈到的很多东西串联起来了。第一点就是,真正重要的可能不是产品本身,在有更偏运营层面的问题时,对体验进行微优化不会有什么实质改变,但如果你是在为用户打造产品,总会有产品的成分在。我觉得这里还有一点很有意思的……其实有两点。一点是它呼应了你关于产品人自主性的重要性的观点——我感觉你当时大概就是,好,这是团队,这是分配给我的任务,然后你说,哦,但这个才是我们真正需要解决的问题,我们围绕它做一个新产品吧。这背后肯定还有一整个争取认同的故事。
还有一点,我们刚刚请了 Uber 的现任 CPO 做客播客,就是这期之前几期录的。那一期全都是关于内部试用的,正好就是关于发现这类问题的。他作为 Uber 司机跑了七八百单来发现这些问题。他有一段话说得特别好——坐在办公室里把司机端 app 做得很漂亮是一回事,但真正以每小时 60 英里的速度开车,手机在几英尺之外,还要在上面操作各种东西,那是完全另一回事。
内部试用与设计思维
Peter Deng: 完全同意。我想起来,我在加入 Uber 之前休了两周假。在那段时间里——我长期以来一直痴迷于用户研究,那个年代如果你真想了解广大用户怎么用你的产品,这尤其重要——我记得我实际上租了一辆车来开 Uber,就是那两周。一辆白色的小型大众什么车来着。我贴上 Uber 标识,打开 app,就开始接单了。没有比亲自内部试用更好的学习方式了。我想接着——Sachin 就是你请来做播客的那位对吧?对,他非常非常厉害。那我就接着他说的往下讲。我觉得对我真正留下深刻印象的框架,是我在学校学到的——因为我接受的是 IDEO 的设计思维训练,我当时在斯坦福的设计学院,那时候我们真的是在拖车里上课,可见当时有多早。
但我记得那个让我印象深刻的框架就是 IDEO 所倡导的:优秀的设计思维有五个阶段。第一是共情,第二是定义,第三是构思,第四是原型,第五是测试。我喜欢这个框架的地方——我真的希望这一点不要被遗忘,因为我不知道现在设计思维教学里还有多大程度在强调这个——是它用了正确的词汇。第一件事是共情。你必须真正感受到客户的痛点。不能只是从理论上理解问题是什么。而是要真正共情,这就是为什么用户研究对我来说一直如此重要——去理解那些痛点,或者像 Sachin 说的那样,亲自去跑那些单,但也要飞到世界各地去。我在 Uber 工作时会去了解各种不同的使用环境。
所以共情是一个非常有力量的词。定义也是一个非常有力量的词,因为它迫你必须把问题表述清楚。这又回到了我之前说的语言的问题——你必须非常审慎地去定义你想要解决的问题。然后构思,我们都知道就是头脑风暴,原型和测试就不言自明了。但我认为前两个阶段非常有洞见,它直接呼应了 Sachin 说的。你必须去内部试用,因为你必须真正去共情,而伟大的产品,正是出现在你真正感受到痛点、真正对人们的体验产生共情的时候。
共情的操作化
Lenny Rachitsky: 你刚才说的让我联想到另一期播客,Linear 的产品负责人 Nan 有一个非常好的概念,和你说的完全一致——作为产品人,你要像你的客户一样感受到他们的痛苦。你不应该停止追问,直到你真正理解他们告诉你的东西,感受到他们所感受的痛苦。基本上,这就是如何把共情落到实处。就是——你有没有感受到那种痛苦?
Peter Deng: 对,我真的希望现在的产品人仍然在做这件事,因为我觉得现在有太多捷径可走,一旦走了,你就会错失关键的东西。我至今清楚地记得,我和 Kevin Systrom 一起飞到洛杉矶做用户研究,那是一个单面镜的观察室,我们听人们谈论 Instagram 以及他们怎么使用 Instagram,没有什么能替代那种体验。我想对那些做了用户访谈然后说”嘿 ChatGPT,帮我总结要点”的人说——你错过关键了。你不可能对一个摘要产生共情。你必须亲临现场,完全沉浸其中,不带手机,真正去听那些话语和语调。只有这样,你才能获得完整的色彩。
Lenny Rachitsky: 这让我想起 Jeff Bezos 有一句很经典的话:如果你有一个个案和一份数据,它们告诉你的东西不一致,相信个案。天哪,今天学到了太多了。好,我们开始收尾吧,聊了很多了。我想快速问一下你关于 Facebook 的事。你很早就加入了 Facebook。我之前提到过的 Eric Antonow 告诉我,你在那个阶段离开 Google 加入 Facebook,是一件很奇怪的事。当时 Google 如日中天,你有着非常好的职业路径,一切都进展顺利,但你决定做一次大的跨越,加入 Facebook。你看到了什么?我觉得这里面有值得学习的东西——你看到了什么,可能帮助其他人决定该去哪里工作。
为什么离开 Google 加入 Facebook
Peter Deng: 我一直对理解我们作为人的本质、理解人性的底层运作方式很着迷。我记得当时和 Facebook 的人交流并了解了他们的想法——那时候人们还说”这不就是个大学网站吗”,当时的风气就是那样。但我看到的是,这个团队以及 Mark 和其他人,真正理解了人们渴望连接、害怕孤独、想要分享这些最根本的人性需求,而且他们对想要解决的问题有非常准确的表述——让世界更加开放和互联。这深深引起了我的共鸣,因为我大学学了很多心理学,我着迷于一个根本问题:我们作为人,底层是怎么运作的?对我来说去 Facebook 工作是不需要犹豫的选择,因为他们理解人性的底层运作方式,知道如何构建与人性相匹配的产品。
他们并不是试图把某种东西硬塞进一个不自然的形态里。他们的思路更像是:我们如何构建技术和产品,来真正增强人们保持联系的根本欲望?这也回到了为什么我认为使命宣言如此重要——你去看看 Friendster 或 MySpace 的使命宣言,我甚至不知道他们有没有使命宣言,或者那些宣言是什么,反正都空洞无物,完全没有触及 Facebook 所追求的那种根本的人性层面,这深深打动了我。我记得当时和 Eric 聊,“嘿,我该怎么办?接受 Facebook 的 offer 还是留在 Google?“但归根结底,是那种与我的价值观的深度共鸣——构建真正符合人性的东西。我认为对任何创业公司、任何做产品的人来说,你构建的东西与人类根本的需求和渴望之间的阻抗匹配(impedance match)做得越好,你就会越成功。
所以这是我的主要回答。第二个回答是,我在职业生涯中一直以学习最大化来优化自己的选择。这是我对很多人说的一个重要原则,因为他们看着我的履历说,“哦,你在所有这些公司待过,秘诀是什么?“我说,我只不过是找到了能学到最多东西的地方。对我来说,那个阶段 Google 已经不是那个地方了,而 Facebook 的运营中有太多我想学的东西。在 Facebook,我在那里待了九年半,但我大约每两年半就会换一次方向,只要我觉得有新的东西可学。
我不知道这算不算什么秘诀,我运气好,有机会学到不同的东西和不同的技能,这对我帮助很大。而且不管结果如何,我觉得这对个人生活来说也是一种很好的方式——就是以学习和体验为优先。对我来说,去 Facebook 就是因为我看到了巨大的学习空间,而最终这些学习也确实发生了,所以我觉得那也是一个很好的结果。
如何选择职业方向
Lenny Rachitsky: 总结一下,给那些正在两份工作之间犹豫、或者在考虑是否该离开去做新事情的人几个要点:第一,你是否觉得自己学到了足够的东西?你想去的新地方是否能让你学到更多?第二,他们正在构建的东西是否与人的行为方式相契合?就是你说的那种阻抗匹配。还有一个你分享的要素是——他们是否对事物的运作方式有真正独特的洞察?以及你是否真正在意这件事?这是否也是你看待世界的方式?你谈到 Facebook 时说,他们对人类行为有非常独特的洞察,而这对你来说非常重要,所以那是一个非常匹配的选择。
Peter Deng: 百分之百同意。关于洞察这一点,谢谢你总结出来,因为这也是我现在寻找的东西,也是我评估是否要与公司和创业公司合作的标准——你有没有独特的洞察?你能不能教我一些我确实不知道的东西?这通常是强烈观点的良好指标,而拥有强烈观点非常重要,因为 Mike 和 Kevin 在 Instagram 时有一句话:“我们可能不是对的,但至少我们不困惑。“我觉得这是一句很美的话,因为有时候你就是得去做你认为正确的事,而犹豫不决恰恰是真正会伤害你、反噬你的东西之一。所以对我来说,无论是作为运营者加入公司时寻找创始人,还是在当前角色中支持的创始人,我都在寻找那些有强烈信念的人。
Lenny Rachitsky: 太有意思了。LinkedIn 的 CPO Tomer Cohen 也经常用这句话,很出名。
Peter Deng: 真的吗?
Lenny Rachitsky: 我觉得他是从那几位那里借来的。对,那是他的座右铭之一。“我们可能不是对的,但我们不困惑。”
Peter Deng: 哇,我不知道这个。我确实和他聊过一次,我不记得有没有聊到这个,不过话说回来,也可能就是英雄所见略同——Mike 和 Kevin,还有 Tomer,不同的厉害的人感受到了同样的共鸣。
失败角(Fail Corner)
Lenny Rachitsky: 我太喜欢这场对话里引用了多少期播客的内容了。好吧,说到学习,在进入我们非常令人期待的闪电问答环节之前,最后一个问题——我要带大家进入失败角(Fail Corner),这和你刚才提到的成长型思维(growth mindset)非常契合。这个环节的初衷是,人们上播客的时候,总是分享各种精彩故事,一切都很顺利,我取得了这么多成功,在这么多了不起的公司工作过,一切都顺风顺水。但现实中,事情往往不会那么顺利。大多数人经历过大量失败的尝试、项目、职业上的挫折,所以问题很简单:你曾经构建并发布过一个完全失败的产品吗?而且我想用你喜欢的方式来问——那次经历如何改变了你的思考和做事方式?
Peter Deng: 一个例子是,既然我们之前聊到了 Instagram,我们当时在 Instagram 尝试做一个相机优先(camera first)的应用,叫做 Bolt,但失败了。它的工艺和设计水准非常高,核心理念基本上是:我们能不能降低分享的压力——打开就是相机,直接把一些东西发给朋友,收到好的反馈,然后从这里出发。而且这显然是 Instagram 的设计团队,所以是一流水准。应用设计得非常好,速度也特别快,因为背后是 Instagram 的工程团队,他们在高性能移动应用方面非常擅长。它拥有我们在 Instagram 所珍视的一切优势,但我们发布之后——我记得是在新西兰还是澳大利亚——就是没能成功。
我记得我们知道这一点的原因,是因为我们在看留存曲线。留存是任何产品最关键的指标,不是用户数量,不是使用量,而是留存,而且是分群的留存(cohorted retention)。你画一条线,如果它能趋于平稳,那就说明你的产品状态不错,因为这意味着用户在一段时间之后会继续留在应用里。但这种情况并没有发生。我觉得这里的教训是,你可以拥有世界上最好的团队、最好的产品品味(product taste),但你无法预测什么会在第一次就成功。
失败是可以的,你只需要从中站起来学习。没有人为此沮丧。实际上,我们当时在 Bolt 里开发了一些技术,后来移植到了主应用上,这非常有帮助。引用伟大的美国诗人 Sean Carter 的话说:“这不是损失,这是教训。“我觉得作为产品人,这一点非常重要——你不把它看作失败,而是看作一件好事。现在我又聪明了一点。这只是我收集的众多例子之一,但我认为这是一个很好的例子,说明了一件有点反直觉的事情:即使你有最好的团队,也不一定能一次又一次地命中靶心。有时候你就是无法预测哪些会成功,你只需要有智慧地说,好吧,让我们看看能从中学到什么,能挽救什么,然后继续前进。
Lenny Rachitsky: 我确实记得那个产品和它的发布,或者说听说过它,但我平时也不会想起它。所以我觉得这是一个很好的提醒。因为 Instagram 发布一个新产品,试图重新定义你使用相机的方式,这可是件大事。所以我能想象它没成功这件事在当时会是多大的事。但与此同时,也没有人真的还记得了。
Peter Deng: 没错,是的。
Lenny Rachitsky: Peter,我们已经聊了两个小时了。我觉得我们还可以再聊两个小时。这部分我们留到下次对话吧。
Peter Deng: 好的。
Lenny Rachitsky: 在进入我们非常令人期待的闪电问答之前,你还有什么想分享的吗?或者想对听众说的,也许是想再强调一下你觉得可能有帮助的观点?如果没有的话,我们就直接开始。
Peter Deng: 我觉得我们直接开始吧,因为我觉得你已经把我那点智慧榨得一滴不剩了。你做得很棒,帮我回忆起了这些故事,重新讲述了这些经历。我准备好了,直接开始吧。
Lenny Rachitsky: 这正是我的目标,虽然我知道还有很多我甚至还没开始挖掘的东西。不过话不多说,我们进入了非常令人期待的闪电问答环节。准备好了吗?
Peter Deng: 准备好了。
闪电问答
Lenny Rachitsky: 第一个问题。你最常推荐给别人的是哪两三本书?
Peter Deng: 这个对我来说很简单。第一本是《Sapiens》(人类简史)。如果你是做产品的,你要为人类构建产品,就必须理解我们自己的人性,这是毫无疑问的。那是一本很美的书。我在它还叫《From Animals to Gods》(从动物到上帝)的时候就读过了,后来换了个书名重新出版,但它一直深深留在我心里。我记得那是一本很短、很容易读的书,所以我非常推荐。第二本书,对于做产品的人来说是一本经典,就是 Don Norman 的《The Design of Everyday Things》(设计心理学)。这本书可能看起来有些过时、有些老,但我向你保证它没有。它真的能帮助你理解实体产品设计,而实体产品同样是塑造和适应人性的东西。我觉得它能给你很好的感觉。
第三本书是我目前正在读的,一位朋友推荐的,我简直放不下来。叫《The Silk Roads》(丝绸之路),Peter Frankopan 写的。基本上这是从丝绸之路和中东的视角来重新讲述历史,以及它是如何演变的。它太迷人了,因为我非常喜欢的一件事,Lenny,就是从不同的视角看问题。这也是旅行有趣的地方,也是用户研究对我来说有趣的地方。它真的帮你用一种不同的方式去看我们都经历过的世界历史事件——我们一直是透过非常西方的视角来看待这些事件的。而且它把很多东西串联起来了:西方思想、东方思想,如果你看到它们之间的联系,会发现非常迷人。我才读了大概两三四章,但绝对是一上手就忍不住推荐的。
Lenny Rachitsky: 你最近最喜欢的一部电影或电视剧是什么?
Peter Deng: 可能不算最近的了,但一直萦绕在我脑海中的是《The Wire》(火线),HBO 的《The Wire》。现在电视剧太多了,我还在消化,要不要把它放进我的历史最佳名单里。但那里的叙事、那些性格一致但又各不相同的角色,以及《The Wire》那种优美的写作,是无与伦比的。
Lenny Rachitsky: 我现在很好奇你的历史最佳名单里都有什么,但我不会追问了。我们继续。你最近发现的最喜欢的产品是什么?
Peter Deng: 我就选 Granola 吧,因为我觉得我们之前聊过这个,但它对我来说已经成为一个超能力。我现在通勤时间比较多,我的做法是进入单人模式——坐上车之后,开始思考和头脑风暴一些投资方面的想法或论点,等我到达目的地的时候,它们就已经以更有条理的方式整理好了,而且有时候连我自己都没想到的表达方式它也帮我组织出来了。它不光是帮你把想法变成文字,还在这个过程中提供了辅助。我觉得它在很多层面上都是一个非常美的产品。
人生格言
Lenny Rachitsky: 下一个问题。你在工作或生活中有没有一个经常回想起的最喜欢的人生格言?
Peter Deng: 有。这其实是我父亲教我的。是一句中文的说法。在中文里它是押韵的,不过在英文里也几乎押韵。英文大概是这样的——树挪死,人挪活。我觉得这句话很有意思,我一直在反复回味。这也回到了为什么对我来说,学习和尝试新体验、在我有幸待过的不同公司工作是如此快乐。我真的认为那才是生活应该有的方式——去体验这些不同的经历。而且这里面有一种诗意:是的,不幸的是,树长大后你没法真的搬动它。但对于人来说,把人搬到不同的地方,我们去不同的地方旅行,去不同的工作,会获得不同的人生体验,我觉得这才是让生活真正值得过下去的东西。
Lenny Rachitsky: 我总是会想我会怎么回答这个问题,有几个答案,但其中一个是我和我妻子决定做某件事时经常想起的——选择冒险。类似的情怀。
从产品领导者到投资人
最后一个问题。你现在从产品领导者转型为投资人了,所以我想给你一个机会告诉大家你在寻找什么样的项目。你已经开始投资初创公司了。你在找什么样的初创公司?谁应该主动联系你——
Peter Deng: 谢谢你给我这个机会。对我来说,答案一直很清楚——我就是喜欢和优秀的人一起工作。投资对我来说,就是能够支持更多杰出创始人的途径。我一直被创始人这种原型所吸引——无论是和 Zach、Travis、Howie,还是 Oculus 的 Brendan,以及 OpenAI 的伙伴们密切合作。我觉得有这样一种令人惊叹的、富有远见的人格类型,我喜欢以各种方式支持他们。过去我主要是作为产品领导者在内部支持他们,但对我来说,核心就是找到那些杰出的创始人。在现在的角色中,我可以同时和许多创始人一起工作。就在两天前,我还和三个不同行业的三位创始人进行了深入的通话和产品讨论,那种感觉让我的思维非常活跃。这就是为什么我现在在做这件事,我也希望能找到更多这样优秀的思维伙伴,以及我能帮到的人。
Lenny Rachitsky: 那关于阶段和市场呢,能不能告诉大家什么样的对你来说是合适的,什么样不合适?
Peter Deng: 当然。我会说早期阶段——种子轮、种子轮+和 A 轮——是我最兴奋的领域。我觉得我能帮助创始人看到下一阶段。我的职业生涯中已经看过太多”电影”了,所以就像——哦,太好了,我完全可以预见这个东西的未来走向。你需要说服我认同那个未来,然后我们就可以一起碰撞,我也能在从一到十、从十到百的规模化过程中提供支持。这一点非常重要。
至于我在寻找什么,就是我之前说的那几点。在当今这个时代,会有非常多了不起的东西被创造出来。第一,你是否有独特的数据,是否有数据飞轮(data flywheel)?第二,你是否有一个真正精雕细琢的工作流程可以深入攻克?第三,我想再加一条——你是否有那种洞察力,知道哪些产品要素真正重要,哪些不重要?然后你如何在此基础上进一步扩展?所以,是的,我非常期待认识更多创始人,不管是通过这里还是其他渠道。
如何联系
Lenny Rachitsky: 好,最后一个问题——如果有人想就这些和你聊聊,怎么联系你?听众可以怎样帮到你?
Peter Deng: 谢谢你的问题。我是一个内向的人,所以我在很多社交媒体上都很安静。我在 X 和 Threads 上都有账号,但 LinkedIn 真的是我选择的社交网络。我希望能被动地浏览和了解正在发生的事情。至于听众怎么帮到我——我就是想学习。你们在想什么?你们看到了哪些洞察?我对当今 AI 的一个比喻是,它是人类发现的一种非常有趣的新元素。而令人惊叹的是,人类本身也非常有创造力。所以人类会用这个新元素做什么,我非常着迷。而且你可以分辨出哪些创始人真正把玩过这个元素,因为他们对这东西能做什么、不能做什么有一种内在的直觉。我就是在寻找被你们所有人创造力所启发的机会。
Lenny Rachitsky: 哇,这个思考角度太棒了。这会稍微改变我对 AI 的看法。Peter,这次对话太精彩了。非常感谢你花时间分享这么多智慧。我知道这是你第一次做这样的访谈。我觉得这会在很多不同的方面帮助到很多人。我觉得我们涵盖了我所有想聊的内容,所以再次感谢你——
Peter Deng: 也谢谢你邀请我。这真的是一种享受,希望外面的一些人能从中学到一些东西,觉得有用。我的目标就是分享一些东西,希望能对一些人有所帮助。所以谢谢,谢谢你给我这个机会。
Lenny Rachitsky: 谢谢你,Peter。大家再见。
术语表
| 原文 | 中文 |
|---|---|
| AGI | AGI(通用人工智能,保留原文缩写) |
| Andrew Chen | Andrew Chen(人名,Uber 增长团队负责人,保留原文) |
| Bolt | Bolt(Instagram 曾推出的相机优先应用,保留原文) |
| cohorted retention | 分群留存(cohorted retention) |
| Copilot | Copilot(微软 GitHub 的 AI 编程助手,保留原文) |
| CPO | CPO(首席产品官,Chief Product Officer,保留原文缩写) |
| Cursor | Cursor(AI 代码编辑器产品名,保留原文) |
| data flywheel | 数据飞轮(data flywheel) |
| design thinking | 设计思维(design thinking) |
| dog food | 内部试用(dog food,指公司员工在产品公开发布前内部使用) |
| Don Norman | Don Norman(人名,保留原文) |
| Dwarkesh | Dwarkesh(人名,播客主持人 Dwarkesh Patel,保留原文) |
| Eric Antonell | Eric Antonell(人名,保留原文) |
| Eric Antonow | Eric Antonow(人名,保留原文;注:前文术语表中为 Eric Antonell,此处保留原文写法) |
| ETA | 预计到达时间(ETA,Estimated Time of Arrival) |
| Fail Corner | 失败角(Fail Corner,播客常设环节) |
| Felicis | Felicis(风投机构名,保留原文) |
| Fidji | Fidji(人名,指 Fidji Simo,保留原文) |
| Friendster | Friendster(社交网站名,保留原文) |
| George Lee | George Lee(人名,Instagram 增长团队负责人,保留原文) |
| Granola | Granola(AI 会议笔记产品名,保留原文) |
| grit | 毅力(grit) |
| growth mindset | 成长型思维(growth mindset) |
| growth PM | 增长产品经理(growth PM) |
| Herbert Clark | Herbert Clark(人名,斯坦福大学心理学教授,保留原文) |
| IDEO | IDEO(设计咨询公司名,保留原文) |
| Ilya | Ilya(人名,指 Ilya Sutskever,保留原文) |
| impedance match | 阻抗匹配(impedance match,借用电子学术语,指产品与用户需求的契合度) |
| Jeff Bezos | 杰夫·贝佐斯(国际知名人物,使用公认中文译名) |
| Jill | Jill(人名,Uber 公关/传播/政策负责人,保留原文) |
| Joanne Jang | Joanne Jang(人名,OpenAI 员工,保留原文) |
| Josh Constine | Josh Constine(人名,保留原文) |
| Kevin | Kevin(指 Kevin Systrom,Instagram 联合创始人,保留原文) |
| Lauryn | Lauryn(人名,Airtable 增长负责人,后转至 Notion,保留原文) |
| Lenny Rachitsky | Lenny Rachitsky(主持人名,保留原文) |
| Linear | Linear(项目管理工具产品名,保留原文) |
| LinkedIn(职场社交平台产品名,保留原文) | |
| LLM | LLM(大型语言模型,Large Language Model,保留原文缩写) |
| managing up | 向上管理(managing up) |
| Marc Benioff | Marc Benioff(人名,Salesforce CEO,保留原文) |
| Mark | Mark(指 Mark Zuckerberg,Facebook 创始人,保留原文) |
| Mike | Mike(指 Mike Krieger,Instagram 联合创始人,保留原文) |
| model designer | 模型设计师(model designer) |
| MVP | MVP(最小可行产品,Minimum Viable Product,保留原文缩写) |
| MySpace | MySpace(社交网站名,保留原文) |
| Nan | Nan(人名,Linear 产品负责人,保留原文) |
| newsfeed | 动态消息(newsfeed) |
| Notion | Notion(产品名,保留原文) |
| NPR | NPR(美国国家公共广播电台,National Public Radio,保留原文缩写) |
| Peter Deng | Peter Deng(人名,保留原文) |
| Peter Frankopan | Peter Frankopan(人名,保留原文) |
| photo tagging | 照片标注(photo tagging) |
| post training | 训练后(post training,AI 模型训练后的微调阶段) |
| PRD | PRD(产品需求文档,Product Requirements Document,保留原文缩写) |
| product craft | 产品工艺(product craft) |
| product taste | 产品品味(product taste) |
| PXD | PXD(Peter Deng 的个人工作风格文档,保留原文) |
| QBasic | QBasic(编程语言,保留原文) |
| Replit | Replit(在线编程平台产品名,保留原文) |
| RPG | RPG(角色扮演游戏,Role-Playing Game,保留原文缩写) |
| Sachin | Sachin(人名,指 Sachin Kansal,Uber 现任 CPO,保留原文) |
| Sean Carter | Sean Carter(人名,即 Jay-Z,保留原文) |
| single player mode | 单人模式(single player mode) |
| South by Southwest | 西南偏南(South by Southwest) |
| The Wire | 《The Wire》(火线,HBO 电视剧名,保留原文) |
| Tomer Cohen | Tomer Cohen(人名,LinkedIn CPO,保留原文) |
| Uber Reserve | Uber Reserve(Uber 预约出行产品名,保留原文) |
| zero shot | 零样本(zero shot,AI 模型在无示例情况下直接执行任务) |
此文档由 AI 分片翻译(translate_long_document)