How to Break into AI PM from Google AI Product Leader | Marily Nika (Google)

The first thing is to get out of your head and not worry that, oh God, I don't have a CS degree, I don't have a PhD, anyone can do it. Now you need to put your learning hat on and you need to be comfortable with just going back and getting uncomfortable and learning about what does each one of these models look like, how do you train and what are the trade offs. So you need to have AI awareness and having that AI awareness should not overwhelm you. Like, you don't need to get deep in the math, but you need to understand how things work and how that affects your PM craft. You also need to just demonstrate to the software engineers and scientists that you're willing to sit there next to them and really understand how things work, really understand the architecture, really understand the technical decisions they need to make. So you lean in them and they lean. You're partnering and they just need to respect you and realize that. Oh, okay. Marilee is a trusted partner. She knows her PMcraft and she's going to help answer these questions because ultimately it's all about the users. Right? Experience.

All right, welcome everyone. So my guest today is Meredith Nika, AI product lead at Google. Meredith got her PhD in machine learning and then she spent 12 years building AI products at Google and Meta. She's also taught over 5,000 students through her top rated AI courses. So welcome.

Thank you. Thank you so much for having me, Peter.

Yeah. So you know, I feel like we got to know each other and why don't we start by talking about just like a quick explainer of some core AI concepts since you do have the PhD and all its experience.

Sounds good. Yeah.

Why don't we start with like, what is your definition of AI and how is it different from machine learning?

Perfect. Well, I love you're getting straight to the important step. So actually I put together this like use cases and applications map. Is it okay if I share my screen as I talk about this?

Okay, go for it.

Awesome. Perfect. So I put together this AI algorithms and applications map just because I'm getting this question so much. And I think this is a really nice way to, to visualize kind of everything that's happening in AI. So as you see, there are different types of learning, right? There's supervised learning, self supervised learning, unsupervised, and then reinforcement learning. I want to start with classification because AI is all about adding intelligence, let's say to a machine, so that it can act and learn in the same manner as a human. So I have a four year old and then we were walking around the other day and I was like, oh, here's squirrel, here's a dog, here is a cat. And then I did that enough times that eventually there was a little animal there. And I'm like, what do you think this is? And my daughter was like, oh yeah, it's a dog. And I said, how do you know it's a dog? And, and, and she came up with some reason. But this is what AI is all about, right? It's learning from data that's labeled specifically in the classification example. And then when we have a new data point that has no labels, it predicts and it provides the low probability as to whether it's like a category doc or. But it's interesting because in this map you can see the kind of applications you can have. For example, spam and not spam, which is like the coolest feature that exists on email is exactly built with that manner, right? So you do classification, you train it smart matching, another fantastic capability that's provided by AI. So, you know, dating apps is a good example on the use case image classification. If you go on Google, Google Photos and then type in, you know, old wedding dresses or sky that's blue, it classifies based on that. So that's supervised learning. You also have a regression. So regression is actually good for forecasting. So you will see that for, you know, player capabilities and how they perform during a game and predicting and guessing that or forecasting sales. That's also machine learning. We also have. Let's go on, let's go on the far left. Unsupervised learning right here. So this is the kind of learning where AI learns without you providing label data. It's more like, hey, here's a bunch of info, you figure it out. And this is where AI does pattern matching. And I think a good example I have here is, you know, clustering, being able to segment images or customer segmentation. Right. This is heavily used by targeted advertisements. It's amazing for anomaly detection and fraud and so on. We also have two more things and I'm almost done, I promise. So reinforcement learning that you see at the bottom, this is where the cool stuff are. Well, cool, like everything is pretty cool. But here's where you will see things like learning through trial and error. So imagine having a dog and teaching the dog how to sit. You give a reward and then it eventually does the right thing and then over time learns the behavior. So this is used, you know, in the real world when an agent interacts with an environment and learns to achieve a goal like DeepMind and of course learning, you know, chess or a robot learning how to navigate in a maze, essentially. I have some cool applications here, like recommender systems. Yeah, you definitely use it there. And then robotics and automation and it's just, you know, I love this map because it's a way to see apps and features you use day to day and how what powers them. And I just loved putting this together. And it took forever. And then last at the very top, we're talking about LLMs and of course LLMs, as folks know, large language models. These enable things like, you know, chatbots and AI agents and content generation and summarization and search and all these things. I know I spoke for too long, but hopefully that was a helpful answer.

So basically a supervised learning is you give it this labeling data where like you have like given this image, given set of images. This is like I'm going to tell you if this is a cat or a dog and then based on this data that I have, you tell me based on another set of images if this is a cat or a dog. Right?

Yeah, but it's, but it's important to know that there's always a probability. So we'll say here's I'm certain by 78% as whether this is a cat or a dog. And that probability is so important for product managers because you build probabilistic products as a product manager, no longer deterministic. And that is a whole other mindset you need to have when you want to bring such products to life.

Makes sense. So do you think LLMs are supervised, unsupervised, or somewhere in between? Like.

Yeah, yeah, I'm going to skip this question. Peter, you remove it, because there's a lot to say there and I don't want some scientists to come to me and say, well, actually, blah, blah, blah, so let's skip this.

Okay? Okay. All right, so that's a really good overview. So now let's talk about AI product management. Right. So there are so many different types of AIPM roles. Maybe you can just give us a general framework for like what traits the best AI PMS have in common.

Oh, that's a very good question.

Yeah.

Well, first of all, let's define AI pms. Right? So whenever you, whenever you work directly or indirectly with this type of models and you have a scientist team and part of your requirements need to be kind of refined and discussed with the, with the science team because there is some model training required, then you know, you're an AI Product manager. And I was an AI product manager before. It was cool. So I started in 2013 and it was kind of like I was working with, you know, scientists and stuff. And I didn't realize that not all PMs do this type of work and have the same type of challenges. So that's when I started to realize, okay, what are the unique traits and characteristics that you need to have in order to be able to do this? Well, number one, you need to be comfortable with experimentation. And when I say experimentation, I mean you will heavily pivot from your original hypothesis and requirements to the final product. Like it's going to be completely different and you need to be okay with that. I know PMs that really need this kind of, you know, step by step approach and roadmap and build against that. But in AI, it's almost never the chance that what you have in mind is what you end up building. Number two, being comfortable with that probabilistic nature. Because as an aipm, you have kind of this responsibility to advocate for AI and educate, if you will, other cross functional partners that are not familiar with it on how it works. I've been in product reviews where, you know, leadership was trying this product we wanted to launch and they're like, hey, I tried this two, three times. And every time I tried it, it had a different output. I don't think it's working as intended. And I was like, it does work as intended and here's why. So understanding kind of this more technical nature and being able to advocate for it is important. And then last but not least, every single day is about trade offs. So when you have this strategic thinking and you're not afraid to take the risk and say, all right, given all these trade offs, here's what I think we're going to do and let's discuss it. And having this strategic, diplomatic kind of approach and discussion, I think is very important. And I feel strategic thinking is a skill that I traditionally used to see in mid to senior levels. Whereas for AI, you gotta have it walking in. So strategic thinking is so important.

I like your point about setting right expectations of leadership, because like you said, this thing is very non deterministic and if you try hard enough, you can always get it to spit out potentially the wrong answer.

Yeah, yeah.

So you have to like, what are some ways to set the expectations upfront that you've seen that are effective?

Good question. And I think this is an indirect question about product reviews, all right. Because you know, in the past there's kind of A forum for product reviews. And you kind of sign up and you go there and you, you say what you want to launch or what the question is that you have and you get a discussion, you can answer, you launch or you answer, you go ahead. But with AI products, and that's my personal opinion, if you go in to a product review and it's the first time this forum or steering committee or leadership, if it's the first time they see what you want to do, then you're lost. Like, this is not what you should do. So when you work with AI products, you need to get constant buy in. You need to always align your thinking. You need to make sure everyone around you, leadership or not, are comfortable with your vision and strategy because there's just so much uncertainty and this can cause unease. And of course we need all these, you know, extra resources for, to build AI products. So you really need everyone to be aligned as much as possible at all times. So I think that's a good way. Communicate often, make sure people have seen what you're doing and, and don't work alone. Like, these are cool products. You should involve as many people as possible. It's a collective effort. So I don't like seeing, you know, in the past I may have seen, you know, PM's working in silos. Like, AI is cool. Let's all get there and work together towards this awesome, cool new thing.

Yeah, like you kind of have to partner with everyone else and maybe even customers and just iterate. Right. Just have to like, yeah, you know, get, get something out there, prototype it and see if it actually works. See if the AI capabilities even support the use case that you're thinking about.

Yeah, and I like that you're mentioning the use case. Right. Because so often in AI you kind of lose sight of what you're doing and why you're doing it. So it's good to always say, hey, there's a problem and a reason behind this crazy, you know, AI system I'm building. So we always need to be able say, okay, there is a use case we're solving for. And this is not just a shiny object because it has AI. So I think it's important to call this out as well.

And can you give us a look into what a day in the life is like as an aipm? I think the term sounds very glamorous, but like a lot of it is maybe cleaning up data or like, what, what is a day in the life like?

Yeah, very nice question. Well, the thing is, I mean, every PM has Cross functional partners. But I find that, you know, with AI there's just so much more unease throughout your cross functional partners and there are even more cross functional partners. So you just meet a lot and you need to run a lot of experiments and be very mindful about, you know, what success means. So you just come up with kind of a roadmap and an idea and you want to get aligned with all these cross functional partners. And sometimes when you have kind of different job ladders you work with, you realize that, you know, they have a different understanding of what success means in AI. So there is a higher sense of responsibility, let's say of and challenge to get alignment. So ton of alignment things, ton of product reviews, ton of thinking outside of the box and then tons of pivoting. And by the way, I don't like the word pivoting. I want to use more evolving. I have a friend called Hans and he's like, oh, we evolve, we don't pivot. I'm like, yes, that's, that's exactly it. So you expect no day to be the same. And alignment and discussions and strategy and evolving.

Is like. When you say alignment, you talk about is one of the challenges, is aligning on shared goals or like what are some of the challenges?

Yeah, yeah. So of course you want to align on the high level goals and why are you doing this in the first place. But when you're deep in the weeds and you are in active model training process, there's a ton of conversations around, you know, what kind of data are you going to use? What is that going to look like? How do you communicate that to the users? Ton of privacy and legal discussions. Right. This is very sensitive. Or we definitely want to get it right. So there's just so many different parties you need to work with in order to make sure folks are comfortable with the work. And what you're building that I feel is just more heightened than classic PMing, if that makes sense.

Got it. You kind of have to be a little bit paranoid, right? Because there's like so many different outputs and so many risks that you have to try to account for.

Yeah. And at the end of the day, like, what is success? And there's a thing called mvq, which is the minimum viable quality. So you know, let's say you have a scientist team and they build together and they build this, this model. Then you have your software engineering team to take that and package it into an experience. So you have UX helping in there and then you test this out and you put it in production. So the quality that you had, that you thought that pure model had is just so different when it's in the, in production, in the experience, and it's kind of like, yeah, all right, what's the minimal viable quality this should have in order for it to launch out in the world? And there's no right or wrong. That's all on you. And of course, getting alignment and it's right. There's just so much responsibility in the AIPM hands that again, there's no boring day. And I just love being in this situation.

Love the ambiguity. Yes. Okay, so like, let's say, you know, I'm, I'm a pm. I'm a growth pm. Or maybe I'm not even a pm. I'm like a product marketer. And AI is the hot thing. I really want to become AI pm. How do I go about breaking into this field?

Just start playing around with AI tools. I mentioned there is AI product management, which is, you know, working with scientists and bringing AI powered experiences to life, but there's also AI for product management, which is, you know, you being the consumer of these AI products that were created by some PM. Right. So just start using them. Use Perplexity, use U.com, use Commodore AI, create your own AI agents, custom GPTs, and just get the confidence needed and just start becoming the expert in your own company as to, you know, how you can best serve your, you know, your, your end users. I keep saying this like, if you want to become a product manager before interviewing, you need to reach the point where every single thing you do, you always need to think about, hey, how would I improve this? So I'm telling people, hey, if you have an upcoming interview and you go get a drink at Starbucks and you use the app, I want you to think subconsciously, how would I improve this app? Always, how do I improve? How do I improve? And AI products are the same. So start using these AI tools, figure out ways to improve them. Figure out how to answer the question, if you were to interview for that app for that company, hey, what would you do next strategically? What would you improve and why? Right? Very important. And, you know, I get a lot of people in my certification saying, okay, well, I did the work. I know that I want to be an aipm. I have this certification to show for what's next. I don't have anything to show for in my resume to get the first interview. And my advice for people is attend hackathons, participate hackathons. There is a wonderful website. It's called devpost and it shows you a list of hackathons all over the world. And the best part is that there is a tab called, you know, like teams and team members and there's a little filter called Looking for a Team. So you'll go in there, you'll submit your profile, you'll say that you're the PM for an AI hackathon and you'll meet people that are developers and UX folks and you all come up with an idea, maybe create a prototype, maybe win, right? There is a lot of perks and money to be won, but it's not about winning, it's about having to show something. And on your resume you can say, hey, participated in five, ten projects and here's the GitHub or here's a website where you can see the prototype and I concepted it and it launched. It's in the app store if you do end up launching or it won. So try to get kind of this experience in any way you can. Another strategy I love is within your own company, Shadow AI PMS. If there are no AIPMs, pitch you to become the AI PM. And don't expect for someone to come and say, yep, you're now named the AIPM for the company. Just start doing the job, work with some engineer, leverage some smart functionality, put a few things together, create a little demo and a little prototype, pitch it to leadership and just become the aipm. Don't ask for permission. We cannot ask for permission in order to achieve anything, right? So become the AIPM that you wish to be. Let's see. And you kind of want to become the, the go to person in your company for AI and AI features and experiences. And this is of something that's claimed, it's not something that you know you're assigned to. So you, you want to be that person. And last but not least, something I tell everyone is you'd be shocked by how many people that I teach in my bootcamp have already worked with AI and they're already AI pms, but they just don't know it. So look into all your past products. Is there anything that has some smart functionality? Because if so, you can tap on that, you can leverage that. If you interview, you can reshape kind of your role a little bit and your regimen. Just make sure to say, hey, launched end to end feature or concepted feature, or created a five year roadmap for something that has AI in its core.

I think the fact of the matter is like no matter what role you're in like AI can probably help you do your job better. So just like learning the tools.

Yeah.

That will help you build credibility. Yeah. Like this morning I spent like so you know, with these interviews I've been trying to get the transcripts to clean up. Right. So this morning I spent, I spent an hour like trying to work with this AI prompt to clean the transcripts automatically before I basically had to hire somebody to do it for me. I finally got the prompt in a place where like it actually works. Like I don't actually have to.

Yeah.

And I actually have to hire a human to, to review it like it seems to work. And that, that was like a really good feeling after spending an hour doing this.

Cloud is amazing. Now you can create apps on cloud and it's been fascinating. You just type in, you're like, do this. And then if you don't like something, like, okay, fix this, do this other thing.

Yeah.

But yeah, there are two things the way I see it, right. There is AI product management, which is what we talked about, and then there's AI for product management, which is the tools that you're going to use on your day to day in order to become better, more productive and just evolve as a pm.

Yeah. I mean obviously if you're an AI product manager, hopefully you're using the tools yourself, including your own product.

Yeah.

Okay, so let's talk about building AI products a little bit more. So obviously there's like a lot of hype for generative AI products. There's a lot of startups out there. But I think the unsaid reality is like a lot of these products have like a huge retention problem. Like I just based on my personal experience, you know, I've tried hundreds of AI products and I pretty much only use cloud and maybe perplexity on a regular basis. So I'm curious if you have any like high level principles or any, any tips for building a successful AI product?

Yeah, I think this goes back to the comment about, hey, are you solving a real problem? Like is there something that you will improve from users day to day life that they will have so much value from that they will just need to keep coming back. I mean there are apps out there that got created just to write the habit and they charge like $2 per person using it and they made a lot of money and that's what they were hoping for. So sometimes, you know, they, the goal is not to have retention. It's more like, okay, let's get everything we can. But you know, in the beginning and I keep saying AI is not a product, right? AI is a tool. So in the beginning it was more like, okay, here's what AI can do for us. Here's a playground here, let's try it out. And people tried out, got excited, so it's natural to lose the hype there. But I feel that by the fact that a lot of people started using it in its raw form, came up with ideas on how to take it and infuse it in experiences, examples. I was using Google Sheets the other day and there was kind of this plugin that lets you opens up a prompt and you're like, oh, I want to create a deck about my new bakery. I want to have like an introduction and this thing. And it just created a deck for me with generated images with the right wording, with the right bio props and everything. So I think AI is about infusing it in current day to day workflows versus having a just a screen where you can type in and prompt. So, and that's, that's the beauty of being an AI product manager, right? To figure out how to enhance the day to day life of people by leveraging AI.

Yeah, I think that's a really important point because a lot of people, when they think about generative AI, they think about, you know, like, like ChatGPT, right? Like you talk to some sort of agent or something. But like you said, I think a lot of the most successful AI features are actually just like removing some friction in some sort of core workflow, like using a spreadsheet or like trying to draft the ER or something. You know, it's like, it's like these like tiny features that are not necessarily like super sexy, but like people just use them all the time.

Yeah. I will say though, there is something else that's super awesome, right? If you have like the prompt and you can talk to it. Okay, cool. But if you change the form factor is super cool. So let me show you what I tried the other day and I'm just fascinated by it. Okay. Hey chatgpt, I'm here with my friend Peter Young and we're having a podcast about AI and its capabilities and how retention drops after people try out AI that doesn't really have an enhancement on people's day to day life. What do you think?

That's a great topic for discussion. Retention often drops when AI doesn't add real value to users daily lives. If the AI feels more like a gimmick than a useful tool, people might try it once out of curiosity, but won't stick with it. To keep users engaged, AI needs to solve a real problem or enhance their experience in a meaningful way. For example, an AI that helps people discover new music they love and connects.

Them with their favorite album. But Right, so what did we do? We used, you know, this like a, what used to be a ui. Right. But we use it in a different form factor and because we were able to speak to it and spoke back in this human like ways, now that's something people will use on a day to day because it solves a different problem, which is the convenience of talking to it and interactivity and conversationality and yeah, I just love using this thing. It's just great.

Yeah, I haven't tried that. That's the latest voice release from ChatGPT, right? I haven't tried it.

Yeah, haven't tried. Has this conversational. So you can interrupt it and it can kind of adjust and say oh, oh, yeah. So it feels like a human. Like my husband, I was asking, hey, who are you talking to about that stuff? I'm like, oh, it's Chatterb.

That's funny because I, I joke with some people that like I talk to some of these AI tools more than I talk to my wife during the day. Like, like I, I, I DM them more than, you know, DM my wife. So yeah, I mean like I've, I've basically come to a point where may, maybe this is not right. But like I basically just, whenever I have a problem, I basically just ask AI, like what's your opinion? And, and most of the time maybe it doesn't have the right answer, but it gives me something useful to build off of. Yeah.

And you know, I. Multimodal is really amazing. Like I had made a post about this. There was like a little bird that was injured in my driveway and I was looking at it, I'm like, oh God, who do I call? What do I do? Should I pick it up? And I took a photo and immediately tragedy was like, oh, this is a pigeon. It has fractured the wing. Do not move it. Here's a number to call based on where you're at and give it some water and crackers. And it was just outstanding. It saved so much time and it probably saved the pigeon's life because the first thing was going to be I was going to pick it up and take it in. But they're like, no, no, no, a professional needs to do that. Outstanding.

Yeah, it's so helpful in everyday situations. I do kind of worry that maybe it'll make us like More lazy and dumb, basically.

Yeah.

But maybe not.

Yeah, I can see lazy for sure. I can see it. Yeah.

I kind of want to just hammer this point home. Right. Like, both of us work at big companies, right? Or, like, have worked at many different big companies. And a lot of times, like, you know, you build like a 1, 2, 3 year strategy document and then you get a bunch of buying and then you. Yeah, and then you build and then you maybe like at some point you ship it to customers, they start getting feedback. But I think, I feel like that process is not going to work with AI products. Right. Or I don't know how you feel, but, like, just kind of reemphasize some of the points that you said earlier.

Yeah, things really move fast and things really change. You need to have some strategy in place, but you also need to reconsider and reassess at all times. And you need to make sure the team has this experimentation mindset and realize that, oh, okay, you know, this is what we're headed towards. But we need to be prepared for some pivot. You need strategy in place. You can just do things without strategy. So that's the first thing that I got asked the other day. And you need to always build for where technology is headed, not where it's at right now. So you make all these assumptions about what we'll be able to do in the future. And, you know, because technology may not be there and resources and all these things, you may need to reassess, but try to think as far ahead as you can in order to minimize the risk of, like, evolving and pivoting. And, you know, the other thing, the other challenge I had found is the team, when they're working in this experimental experimentation environment, they're like, wait, so how is, you know, progress assessed? And how is my career being assessed with products if we don't launch? You know, the traditional product management is like, launch, launch, launch. But if you're in AI, it's more like, what question did I answer? And what an. What question did I answer? What input did I get? One insight. How can I pivot? So it's more a milestone basis rather than feature launch in zero to one product. So gotta be mindful about that, that, you know, assessment changes in AI product management as well.

Okay, so like, you. You want to, like, the foundational models are getting better all the time, right? So you want to build your product in a way that leverages the fact that this stuff's sort of getting better, like every month or even every every week. Yeah, that's one point, right?

Yeah.

And I guess your second point is some of this stuff, like, might not make it past like the prototype or notebook stage. And like, that is also success if you figure out early that this use case isn't working right now. Yeah, yeah.

When I was doing my PhD, my first experiment, that just didn't work and it didn't have any. Like, I didn't prove my hypothesis and it was just, oh my God, I lost like three, four months of my, my work. And my, my amazing supervisor was like, this is absolutely progress. You can publish it in the conferences and say, hey, here's what I tried, here's what didn't work, and here's what I would do next. And I'm telling you this to support you. And that paper got accepted and it ended up being like a big chapter of my PhD. So it's an important point that in AI, learning is success and also for product managers, like, it's very important.

Yeah, yeah. So sometimes in product management, you try to structure experiments so that like, you know, like, it's guaranteed to work. And I'm gonna get my next promotion. But I think in this case, you really have to keep an open mind.

Yeah, it depends. Yeah.

You know, now that you work at Google, I'm sure you can use like Gemini in house all you want. Like, how do you use AI to do PM work? Like, which parts of PM work can you automate through AI? Cannot.

So when I'm talking about a zero to one product, you know, there's a lot of hypothesis making and, you know, I may test my hypothesis and say, hey, I think, you know, when it comes to maximizing the. This is just an example like streaming time or, you know, someone watch time on, on, you know, a music app or something like this. I believe that this and this and these are factors that affect this decision. And it kind of double checks my intuition and it gives me extra data or it can quantify things like, hey, here is the bigger sugars assignment than this. And. But also for the implementation side of things as well, I may have a product review and I may say, hey, here's my goal. Here's the problem cons. Can you put together one seater for me to. Just so that I can go in for the review. And it saves so many hours because you take that, you paste it, and then you can kind of, you know, tweak it and adjust it the way you like it. And it's kind of wonderful how much time it saves from the, you know, the in tasks now, of course you shouldn't share anything confidential or anything like that to these apps, but keeping at high level has changed my life, to be honest.

Yeah, I think it's much easier to work off a first draft than like a blank page. Right. So, yeah, getting it to. I've also found it very helpful for, like, summarizing stuff that's like one of the best use cases that it's good at. Like, you know, like, like a 50 message Slack thread like that I don't really want to read. Like, just. Yeah, yeah, summarize for me.

Yeah, yeah. I was. I was interviewing somewhere at some point and I think I had taken the transcript of the entire investor relations release that was like four hours long on YouTube and they uploaded it and I'm like, hey, I'm interviewing for this role. What of all this matters? Summarize it for me. And it was like, oh, okay, cool. And then I went to the interview and I was like, hey, oh, by the way, I read this and there was this and there was that and like, yep, perfect. So it's kind of, you know, changing life in so many ways. Ways in, you know, productivity, let's say.

Yeah, it saves a lot of time on that. That's just kind of. Do. Do you also, like, use it to write PRDs and stuff or, like, do you use.

Yeah, I have a custom. I have a custom GBT that I can share with you. It's called AI product GPT and it actually creates PRDs for me. And it just really, as I said, like, saves time. So it's this one. You can say, I am a PM for Netflix and want to improve retention for binge watchers. Yeah. And then I train it and it kind of comes up with ideas that are AI enhanced and it tells you, hey, well, these are three ideas. Which one do you want to go with? And you choose there like AI powered content recommendation with a solution. The problem solved, dynamic episode interruption. That's cool. Smart notifications. And then it's like, which one do you want? They'll say, let's go, it's free. And, you know, it will take as an input a template I have provided for an AI prd. And then it will just start doing the framing and the market and the pain points and the RICE framework. But the best part of what it does is it generates a mock which actually you can give to the designers and be done. So it's just super, super fantastic, like this type of work.

Oh, wow. You're actually able to get it to.

Generate a mock yes, yes. Check this out.

Wow.

Boom. It's just been fantastic. And I can see the data of how many people have used and it's like, I don't know, like 10,000 or something like that. There you go.

Wow. Well, that is fantastic. Yeah, I never thought about the mock part. The mark part helps.

I mean, the designer will probably be the logo. And also it has become so much better from the hallucination side because now text renders. Okay, like still has some issues. But this is so much better than it used to be. Snares.

Yeah, that's really smart. That's really smart that it generates a mock.

So, yeah, long story short, yes, I use it for PRD generation.

I mean, PRD is like also keeping the PRD updated is like a huge pain. So, like, you know, if you can just use AI, like people have discussion about the PRD and they just use AI, be like, hey, can you update the PRD based on this discussion? I guess I'm admitting how lazy I am. But, you know, you still have to audit what it does. But it saves a lot of time.

Well, not just that. I wish you could like talk, see the comments that people leave and address the comments too. I think that's where the human intervention comes in. But yeah, maybe in the future.

And what is your favorite use case for AI in your life outside of work?

I love this voice command. You know what I like using that's recent? I like generating music and there's this app that does it. And you know, I absolutely love Coldplay. Coldplay is kind of my thing. And I was like, hey, play, generate Coldplay. And it actually did generate Coldplay liking songs now, you know, it has some license data. And then I realized that app had some unlicensed data, so I stopped using it. But I think AI music is something that we are going to start seeing in the future because there's just so much potential in it.

Is this Suno or what did you use for music?

Yeah, that's the one. Yep.

Yeah, yeah, yeah.

It's.

It's amazing. But you know, like, between you and I guess whoever's listening to this, like, I'm pretty sure all this stuff is trained on licensed data. I don't see how it cannot be, so.

I hope so. I hope so. Yeah. I don't know. I don't know. I need to do more research. Yeah.

Yeah. Cool. All right. All right, so let's wrap up. Why don't you talk a little bit about your AI courses and what you cover your AI courses and where people can find out more.

Yeah, sounds good. So, yeah, I really wanted people to break into AI product management and I created an AI product building AI products 101 course on maven. And that was like the top rated, best course only course on Maven that taught people the AI product life cycle and what your role is as an aipm. And all these algorithms were kind of discussed from the PM perspective as well as a ton of exercises and a live coding workshop with no code tools. And people really liked that. And then after they were graduating they were like, hey, we need more. So I created an advanced AI product management course where I bring people from the industry like VP from Microsoft's OpenAI and all these places and they present their exclusive kind of case studies of what it is they worked on. And it's more for like directors, VPs, C level folks and you know, they learn how to create a ipmorgs, they learn how it's applied in other companies and they do all these exercises as well. And then, you know, that was another very, very popular course. So I decided to create a 12 week course that was more like a bootcamp AI product bootcamp that includes the one on one, the advanced and the Capstone project where people can apply their skills to their real world problems and also pitch to a panel of VCs I'm bringing in. And people have raised funding, people have met fantastic colleagues. We have an offline community of 5,000 people that people refer each other, people hire each other, people from startups with each other, we do meetups and yeah, it's been like best rated course. People have been loving it and I, I found my identity as an AI educator and it's, it's been something I absolutely love doing so. And I'm also launching a new course soon. So if anyone is interested in learning more, I'm on LinkedIn, I'm on Sapsack, and yeah, I'm just very excited to show more about this.

Cool. Awesome. All right, well Meredith, thanks for conversation. Thanks for sharing some of your insights and hopefully we get more AIPMs.

That's my mission. That's my mission. Yeah, absolutely. Thank you for having me, Peter. That has been super great.

All right, cool.

All right, thank you.