# Becoming an AI PM | Aman Khan ! [ rw-book-cover] (https://wsrv.nl/?url=https%3A%2F%2Fsubstackcdn.com%2Ffeed%2Fpodcast%2F10845%2F6f98962f26f1b886bac3a845bb7ee00b.jpg&w=100&h=100) URL: https://share.snipd.com/episode/dacce24a-320e-447d-aabf-f8e14459ee0e Author: Lenny's Podcast: Product | Growth | Career  ## AI-Generated Summary None ## Highlights > **3 Types of AI PMs** > - AI Platform PMs build tools for AI engineers, like observability platforms. > - AI Product PMs focus on products where AI is the core experience, such as ChatGPT. > - AI-powered PMs leverage AI to enhance existing product management tasks. > Transcript: > Aman Khan > I think on top of that, there's also AI product PMs. And AI product PMs are really the flavor of PM where the core product is centered around AI. And an example of that would actually be something like a chat GPT. So the core kind of tech or actually, another good example would be, you know, you had Ryzen and Notebook LM is a really great example of that too. And what I mean by that is the core experience is enabled by the model underneath the hood. That's sort of the secret sauce. It's like, you know, you have these researchers and engineers really pushing the boundaries of technology forward in AI. And then your job is really to package that technology and make it consumable for either a business or enterprise or a consumer to really utilize that tech in some way. AI-powered PM role, where the core competency, the core tech is really what you're packaging and distributing. I think the third type of PM here is really the one we were talking about a little bit earlier, which is this concept of you're an AI-powered PM. And to that end, I think that you're going to be enabled by AI technology across the entire stack of the role you might already be doing as a product manager. And we'll get into like a little bit what that means as well. I have some ideas and thoughts around that. But really what an AI-powered product manager in my mind does is you're not really responsible for building the core model from the ground up. Maybe you don't have the access the resources or research teams that you might have at like ([Time 0:08:27](https://share.snipd.com/snip/0b05b676-72a9-4135-a018-1929142498f2)) > **AI PM Tooling** > - Use Cursor and Replit to quickly build functional prototypes. > - Use Vercel's v0 to generate landing pages and basic UI from prompts. > Transcript: > Aman Khan > Obsessed with Cursor. I'm obsessed with Replit. These are tools I use all the time to just really build a prototype and try to understand myself what's possible. That kind of is super enabling for a product manager to show up with a functional prototype. It's not going to be the thing you go and ship in production, but at least the buttons work, and maybe it's like wired up, you know, reasonably correctly to tell a story. There's a really great tool. A lot of folks might be familiar with Vercel if you're trying to create a landing page. ([Time 0:11:06](https://share.snipd.com/snip/b521d61b-6610-4a8a-99a3-72c7a6b2ad44)) > **Getting into AI PM** > - First, build a foundation in machine learning and AI by learning the basics. > - Then, explore your interests and apply that knowledge to problems you want to solve or industries you're passionate about. > Transcript: > Aman Khan > And then how do I stand out as an applicant? I kind of feel like these are maybe two distinct things. You kind of want to start with a foundation of what is machine learning, what is AI, what's the knowledge that has really powered this evolution in technology. And you kind of want to keep driving in that direction with what are your interests. So if there's really two dimensions, you're trying to understand, okay, what are the things I need to know? And then how can I use that in the problems I'm trying to solve or the industries that interest me? So I think there's like this propensity of you want ([Time 0:19:02](https://share.snipd.com/snip/81c222df-dc81-4fce-bf04-3f73c5298bcc)) > **Standing Out as an AI PM Applicant** > - Build a portfolio of AI product prototypes to demonstrate your interest and skills. > - This showcases your ability and excitement to a hiring manager, answering key questions before an interview. > Transcript: > Aman Khan > Maybe to bring this back to an earlier point, there is no program, you know, there's no college degree on product management. So there definitely isn't a college degree on AI product management. And so I think the way that you stand out is through showing interest in the space and building your own skill set and skill stack around what is the foundation of AI and how do I apply it To the industry I'm interested in? And that's actually related to the second point of how do I stand out as an applicant? And it kind of feels like this moment where you can actually go and have almost like a portfolio of products that you've tried to build and maybe they're just prototypes, but they really Help you stand out to a hiring manager that might be deciding, you know, who's the person I want to bring on board. And maybe like, you know, to zoom out for a sec, as someone who's been a part of the like hiring process for AI PMs before, I think the hiring process is really designed to understand sort Of three things about a person. And that's, can this person do the job I'm hiring them to do? Are they excited to do the work that we do here? And do I like to work with them? And so I ([Time 0:20:17](https://share.snipd.com/snip/74aee360-c0fc-4b3b-92ea-8b0f5cba46ba)) > **PMs Thrive in AI World** > - With the rise of AI tools that excel at building, the hardest part becomes knowing *what* to build. > - PMs are uniquely positioned to thrive as they identify problems, articulate solutions to AI, and possess the taste to discern what works in the market. > Transcript: > Lenny Rachitsky > Really happy you shared that. And that's reminds me, I'm just going to go on a tangent here. I want to get your take on this. I had this kind of hot take a month ago or so that there's this sense that as AI tooling emerged, that product managers are screwed. Why do you need product managers when you could just build things? Engineers could use all these tools, designers, like what's the point of a PM in a world where AI can just build things for you? And I've realized it's exactly the opposite, that you may not need any other function if you have a PM, because if you think about it, what are these AI tools amazing at? They're amazing at building things. You tell them what to build, they build them. The hardest part becomes knowing what to build, finding opportunities, problems people need solved, and then articulating it very clearly to an AI tool what to build, and then having The taste and sense of what is good and great and what will work in the market. And that's like exactly ([Time 0:25:42](https://share.snipd.com/snip/2ca4b266-41c5-4787-b9f6-f26212996461)) > **Avoid Copying Foundational Models** > - Don't simply copy popular AI interfaces like chatbots just because they're familiar. > - Instead, focus on solving real business problems and finding the right interface for your specific AI product. > Transcript: > Aman Khan > Was that the right thing to do? So coming back to your earlier question, how do you kind of go and thrive as a top 5% AI PM versus maybe an AI PM that isn't as stellar? I actually think it's kind of coming back to that earlier point of don't do what everyone else is doing. Just because it feels intuitive to go and replicate this interface that feels really great and feels everyone's bought in because they're familiar with it. You really have to think, is this the thing our business needs right now? Is this the problem that we need to go and solve? And I think if you look deeper at that, you'll actually find that the interface for AI might look really different. It won't necessarily look like a chatbot. It might look like you're trying to optimize or speed up a part of the process that humans have to do today, that you can make their life easier. And for us, that was data analysis. We have to look at a lot of data to make decisions within our own company. And so we decided to try to automate that. There is sort of, you know, we did also ship a small chatbot just to understand how the tech work. But on top of that, our interface around AI looks so, so different. And I think that a lot of the cutting edge sort of experiences around AI kind of look like that too. They don't really always resemble chatbots. And I think that that's really the question you have to ask yourself is, what's the right interface when I'm designing a product like this? > Lenny Rachitsky > So what I'm hearing is if your team is building something that's very similar to some other, to a foundational models product, for example, probably a red flag, probably not going to Be a huge opportunity. So optimized for things that feel different than new. Anything else you want to add there? Because I have a question. > Aman Khan > Yeah, I think there's, there's like, so that that's, that's almost like a point of like, what do people do for last few years? And now there's this moment of this actually kind of happening again. And this moment is sort of happening around the term of ([Time 0:31:23](https://share.snipd.com/snip/38a5bb44-c102-4ee8-804d-adcafc331028)) > **Give Users Control** > - Betty Crocker's cake mix initially failed because it only required adding water, taking away customer control. > - Adding eggs back to the instructions increased sales, demonstrating the IKEA effect where users value having influence over the outcome. > Transcript: > Aman Khan > Job away entirely. And there's, there's this story of, you know, Betty Crocker, like actually, you know, launching this, this cake mix, where all you have to do is like mix water. And they all you had to do was mix water and you could kind of bake a cake with just like one ingredient. All you needed was this cake mix. And it turns out that actually tanked their sales. So all you had to do is basically change the instruction so that all you had to do was add eggs. And that one thing kind of created this effect where their sales skyrocketed because they realized that the customer actually wants to have a little bit of control over the experience. And you can kind of see that even now if you've been able to take a Waymo or a self-driving car. Like the car is fully self-driving, but you can still control the air conditioning. You can still control the music. And so you want to kind of leave some knobs and levers for your customer to feel like they're still in control of the experience. And that's kind of like the IKEA effect applied to AI. Like people feel more empowered when they actually have an impact on the end experience versus everything being taken care of. So that's just like one insight we realized from looking at successful AI products versus ones that try to automate everything away. And I think if you're like looking at the space, you're going to find so many more examples like that. That's such an important point that even if you can build a product that is fully automated, you may not want to. > Lenny Rachitsky > And that Betty Crocker example is really interesting. It reminds me of like Blue Apron and all these food making. I have a friend who told me that he cooks, he's cooking a lot these days and just like finding all this time to cook. And then it turns out he's using Blue Apron. And it feels to you like you're cooking. And you can tell people, hey, I'm cooking a lot. > Aman Khan > But it's not quite the same. Yeah. Actually, if you were mine, I could probe on that. That's a really interesting point with Blue Apron. Because if you look at that product, what's the problem that they're trying to solve? ([Time 0:40:09](https://share.snipd.com/snip/fec972ef-dde1-4fca-8d08-f0f62b7b660e))