“I’m incapable of doing my job without AI”: How this top PM uses Claude + ChatGPT as his second brain

0:00You've actually set up these independent, fully loaded brains in Claude and ChatGPT and then use those to hold the context for you and be more efficient with your job. So show us what one of these brains actually looks like.

0:14Amir Klein:Yeah, let's dive right into it. So here's an example of a project. You can see here that I have a bunch of them and in it we have my files and my instructions. Right now there's 20 files. It started with two or three or four and then just kept growing and growing in terms of its brain. This is what all the projects look like. They all just have a lot of files, instructions, and you can see my endless amount of threads that I have with it. Okay, so what do you think are important files for a product manager to put in? So I usually kick it off with a ping pong. So I give it any data that I have.

0:46Amir Klein:If there is anything that references as to what exactly is going on to start things off, then it helps me and the thread that I'm having here in projects get an understanding as to what I'm looking to achieve and more or less like what we're talking about.

1:02Welcome back to How I AI. I'm Clara Vo, product leader and AI obsessive here on a mission to help you build better with these new tools. Today I have AI PM Amir Quine who works at monday.com and is in the top five AI powered PMs on his team. He's going to show us how he uses AI to build scrapers to get customer feedback into his custom GPTs, use those custom GPTs to not only make his product sense better, but improve feedback that he's gotten from his boss. And finally, he is the first guest that will demo voice mode live for us on air to show how he prepares for product interviews. Let's get to it.

1:42This episode is brought to you by GoFundMe Giving Funds, the zero fee DAF. I want to tell you about a new product GoFundMe has launched called Giving Funds, a smarter, easier way to give, especially during tax season, which is basically here. GoFundMe Giving Funds is the DAF, or donor advised fund, from the world's number one giving platform, trusted by 200 million people. It's basically your own mini foundation without the lawyers or admin costs. You contribute money or appreciated assets, get the tax deduction right away, potentially reduce capital gains, and then decide later where to donate from 1.4 million nonprofits.

2:24There are zero admin or asset fees, and while the money sits there, you can invest and grow it tax-free so you have more to give later, all from one simple hub with one clean tax receipt. Lock in your deduction now and decide where to give later. Perfect for tax season. Join the GoFundMe community of$200 million and start saving money on your tax bill, all while helping the causes you care about the most. Start your giving fund today in just minutes at gofundme.com slash howiai. We'll even cover the DAF pay fees if you transfer your existing DAF over. That's GoFundMe.com slash HowIAI to start your giving fund.

3:12Hey, Amir. Happy to have you here.

3:14Amir Klein:Thank you, Claire. Happy to be here. So one of the reasons why I'm excited to have you here is you don't just like using AI. We were just talking about it. You are now highly dependent and maybe incapable of doing your job without AI. So tell me how you got to this point where you're really depending on AI to do your job as an AI PM. Yeah, I think there's like the whole context switching is something that's really difficult for everyone. And as a PM, when you're going back to back meetings and jumping between different initiatives and different requirements and asks from people, You have so many things to juggle that I discovered with the help of everything that's out there online that you can silo like brains of conversations and threads in GPT and Cloud that it made me super dependent on it in order to just do my work.

4:07Amir Klein:And I'm doing it really well. I think I'm doing it really efficiently. So that's how I got to where I am today. Okay, so what you're saying is you, like all PMs, have to context switch and hold a million different threads in your mind at one time. And instead of burdening your poor human brain with all that context switching, you've actually set up these like independent, fully loaded brains in Claude and ChatGPT and then use those to hold the context for you and be more efficient with your job. Absolutely. Okay. So show us what one of these brains actually looks like. What are some of the use cases for my collection, my file folder of brains?

4:51Amir Klein:Yeah, let's dive right into it. So I'm going to, I guess, dive into what a, I guess, folder looks like, my projects look like. So here's an example of a project. You can see here that I have a bunch of them. And in it, we have like my files and my instructions. So right now there's 20 files. it started with two or three or four and then just kept growing and growing in terms of its brain and i'll dive into like some of the files that are in here and kind of like what i'm trying to achieve to help you understand like how i put this into action but just this is what all the projects look like they all just have a lot of files instructions and you can see like my endless amount of threads that i have with it um like non-stop okay so what do you think are important files for a product manager to put in?

5:42Or what are some unique sources of, you know, context for these GPT projects that you have found particularly useful or unique?

5:50Amir Klein:It varies, obviously, on what you're working on. But I always start with some sort of like reference that it can understand whatever it is that I'm starting with. So I usually kick it off with a ping pong. So I give it any data that I have, any data whatsoever. So if there's a company kickoff deck, if there's a team kickoff deck, if there is another reference to a PRD, if there is anything that references as to what exactly is going on to start things off, then it helps like me and the thread that I'm having here in projects get an understanding as to what I'm looking to achieve and more or less like what we're talking about.

6:25Great. And then what do you want these, you know, second brains or these GPTs to actually do for you? So what is the, you know, what is this one actually do for you?

6:34Amir Klein:I'll start this one kind of like where it began. I started at my company at monday.com leading like the AI agents team the whole initiative there and I'm sure anyone who's on AI today can allude to the fact that AI agents is probably like one of the most fantasized thing that is in SaaS. Everybody wants it. Everybody wants it to work. Everybody has like this vision of what it can do and it can do everything and whatever you hear outside is also what you hear inside and there's a thousand different voices and different opinions of where they want it to go and what they believe it will be. And I was seeking some voice of reason that was unbiased.

7:16Amir Klein:I was looking for conversations that people were having that weren't connected to my company, that didn't have any specific narrative they wanted to go down. I just wanted to know what people are interested in, what they expect, and what they want, which is what led me to Reddit. Okay, so you were a PM. You were given the great and wonderful task of building out agents. No one has agreed on what the definition of an agent is. No one has agreed what the agent should do. No one really understands if customers actually even want agents. technically there's a lot of complexity and aspirationally there's a lot of goals around these so this is a very common product manager problem where you're given a big meaty interesting technically complex problem with a lot of hopes and dreams in it and then it's your job to go make it reality and so what I'm hearing from you is you went and sought out these different resources that were external that could give you more of an unbiased filter on which to make some product decisions and maybe even go kind of back to your leadership partners or people with big hopes and dreams and say, this is what people actually want or this is what's actually possible.

8:29And so you mentioned Reddit. What did you actually pull from Reddit that you think was so useful here?

8:35Amir Klein:So I'm going to walk you through like what I did here at Nora Descrape Conversations on Reddit with the help of Claude. I had this idea, I had this vision that I'm going to be able to get this file or many files containing thousands and thousands of conversations that people have about anything that I wanted. In this case, I wanted about agents, AI, money.com, all in relation with one another. So I went into, I went into, to cloud and I said, here, help me, help me scrape conversations. I started my conversation here with just like, I want to find everything that's written about money.com, whether it be on Reddit, Twitter, LinkedIn, anything that's online um and i wanted like a sort of an automated free way to do this so it went into like a bunch of different options we realized that linkedin and twitter have like more limited apis some of them are like paid eventually we got down to what i needed how to get it and how can do more automated so i asked okay great we know reddit is the place i need to be now give me a step-by-step guide of what i need to do so i'm not not technical i am a little bit i have a little experience with things.

9:39Amir Klein:But I wanted to make sure I'm not skipping any steps. I want to make sure that everything I'm doing here is going to be exactly what needs to be done because I had a vision of what I'm getting out of this here. So it literally walks me through here. Like, open Safari, go to brew, write a command, open your terminal, write, like, literally step by step what I need to do as if I'm five years old and I don't know how to program or code or anything, which was fantastic. It tells me to go into Reddit, go into this, make a developer account, literally follow step by step. I didn't even try changing any name.

10:08Amir Klein:I limited it as it asked. And then we get to, you know, doing brew install, it's getting stuck, whatever. Eventually, we get to the good stuff where I have everything set up. It's finished. Now what do I do? I had the environment set up and I had all the packages that I need to set up. And now it starts giving me like all these commands to run. So install different packages and libraries that I need in order to run the script that I write for me. Here's a script that I'm going to help you write. And let's get started. So basically, it writes me up this script we can see over here where I put in my client ID it would ask me to take out from Reddit, my client secret, and my username.

10:49Amir Klein:And then it scrapes everything that has to do with Monday, with things that I'm looking for with AI and whatever it be. And lo and behold, came out a file with 34 ,000 rows of conversations. Okay. So you have this like very, again, a very classic product manager problem of, I know people are talking about this product. Like I know, I know there's information out there and I can manually call through and, you know, do Reddit searches and search through Twitter, but I want to do this in a programmatic way. I'm technical, but not that technical and dear God, I don't want to write a bunch of Python.

11:23And so what you did is you went to clot and you said, I have this general problem. I just want to know what's out there. You actually said the prompt is really interesting. It says, I want to know everything out there, everything everyone is saying. That's a very broad, yeah, very ambitious flow. And then, you know, Claude narrowed you into, well, here are the actual accessible sources of data for you. And then you wrote a script to go pull all that Reddit data. And then what did you do with that? You know, you said 34 ,000 or something, the 3400 line spreadsheet. What did you do with that data?

12:01Amir Klein:So I'll show you. Basically, first of all, I want to show you what the file looks like just so you get a glimpse as to what this like monstrous file is. So this is what 34 ,000 rows of threads looks like in relation to the conversations that I wanted, the theme that I was looking for. So this is great. And it really helps me understand it has a lot of conversations. I also went crazy with it. I said, like, give me different files, Show me how it looks like in terms of our competitors. Show me conversations that are happening in relation to our competitors, like comparing us to any other competitor and their AI and give me some sort of idea.

12:35Amir Klein:When I had all that information, I went back into Cloud. And I said, okay, now be my analyst and start breaking this down. I'm really into data. I have a little bit of a data background. So I came into it and I said, here, I'm giving you this file. This is one of the files that was pulled out with 30 ,000 rows of conversations. And I said, I want you to summarize it in a table where you put like the frequency. So how often it comes up at what percentage I need weights because I need numbers to go back to my team and for myself to know what to prioritize, see what the biggest, hottest topics are that people are discussing and some of the key discussion points.

13:14Amir Klein:And out comes this table, which is really nice and really helpful because I have, OK, here's here's here's some sort of idea. I know like what are the top five areas that people are wishing and hoping and having conversations about around this, their expectation and their dreams around what AI or an agent would do. And did you find that this did you spot check this at all? Did you just trust it? Did you feel like this was relatively accurate based on what you saw in that in that spreadsheet? Yeah, it's a really good question. I did check it. Obviously, I'm not checking 34 ,000 rows of conversations, but I did do like keyword searches and I looked at different threads and conversations that I put up.

13:50Amir Klein:Also, what I typically do with whenever I ask GPT or Claude to analyze something, I say reference. Give me at least one or two quotes as to where this came up from. And I have tons of conversations where it does that. So then I fact check it. I copy that. I go back into the document and make sure that I was actually there. Awesome. And so you built this scraper. You took the data. You analyzed it. Did you then feed this analysis back into that GPT project we saw before? How does this become part of your corpus of one of your secondary brains? Yeah, fantastic. So I'll show you how that looks like back in the project itself.

14:28Amir Klein:So here I am in the project. You can see that I have the AI threads on Reddit. And in general, these are different Reddit files that I had. And this is how I started things off. I had my kickoff that I had that my manager did. And I had a couple of pages that like I took from our site of like how Monday works. And I just command it. So it's a PDF and put in another classic example of everything is a text. And then so it has an idea. It knows what Monday is. It knows what in more or less like what our goal is as a team. And here are some conversations that are happening. So you have some sort of idea as to what people's dreams and hopes are when it comes to AI.

15:09Yeah, I just want to call out that little trick for folks that might have missed it because I do the exact same thing, which is I actually just use our marketing website or our support website and I just print things as PDFs and download them and then upload them to GPT projects or whatever kind of store because then I know it's scoped to that specific piece of content. I've done it a lot with like support pages. I've done it with the homepage, obviously, because that's usually like the source of truth of your positioning. And then I do it a lot with like pricing pages. So I'll go like print a bunch of competitor pricing pages and drop them in a GPT.

15:40So I think that's a really great source of information for a GPT?

15:44Amir Klein:No, for sure. I think like once you like when I started with this and I had the CSV file from from Reddit, my mind was like, OK, literally everything is text. So like even you may not necessarily think of like a slide. It was a slide from like Google Slides as like text, but it is. So you just download as a PDF and you have just like whatever it is, however many slides here as a reference um and like the example you said as well like the list is endless of what is text so i have it all for like instructions more or less of like what i want to do uh make sure that it is a professional and product management expert and all this stuff that has a good product strategy mindset and product sense mindset um i also give it like a lot of like instructions on you know you know how to give feedback to me you know i did candid feedback and you're not going to be like too nice to me, you know, to challenge.

16:41Amir Klein:It's because I hate it when I will be like super supportive of every idea that I come up with, like push back on things. You have the knowledge. So feel free to challenge me on things I come up with. And then the ping pong begins. And then we start going back and forth and have thousands of conversations here. And so are you using this as a thought partner? Are you using this to prep for meetings with your boss? Like what are the most common use cases for this GPT for you? All my projects, I start with a scope. I want to get to some sort of like outline overview that'll help me with PPRD, with a product review, something that I can deliver to my team.

17:14Amir Klein:So it always starts off with a ping pong, always. And then once I've gone to some sort of document and I have like this overview doc of what it is, I played around with it. I've downloaded it and I put it back into its knowledge. And now it has that as a basis. And then it goes from there. The more work I do with it, the more knowledge I add into it, into its files. and then everything that comes after that is just super intuitive and fluid. I'll give you examples like if I need to write something, how often I get asked questions by like marketing and with the communications theme of like, okay, you're releasing something.

17:48Amir Klein:Can you give me like a two-line sentence is what it is? Those are classic examples of like catch me in the middle of the day on Slack and ask me then and I don't know. I don't have the capacity to think about that. But here is something that has a brain for this. I ask that question and usually does a pretty good job. And then lo and behold, here we go. Great. And then, you know, speaking of documents, I think one of the common complaints that, you know, people have about AI generated documents is the writing maybe isn't great. And at the very least, maybe it's good thinking, but it's quite long.

18:20And I know you've also built a custom GPT to make sure that your writing is concise. So can you show us a little bit of what that looks like?

18:27Amir Klein:yeah there's all also a background to that um so if any of my past managers and colleagues see this um they will deal with to confirm this but i've been giving feedback multiple times and my writing is good but it can be really long so my slack messages on updates and channels are really long and it's hard for you to get through and lenny released i think there was a newsletter where you had Wes Cal, like, say, more or less, like, her recommendations on how to write concisely, how to deliver a message well. And again, example here of everything is written. I took these newsletters, I command feed it, downloaded them, and I put it into a custom GPT as, like, my guidelines to how to write.

19:10Amir Klein:And I'll show you what this custom GPT looks like. You've seen the doom and gloom headlines. AI is coming for your job. But the reality is a little bit brighter. In Miro's latest survey, 76 % of people say AI can boost their work. It's just that 54 % still don't know when to use it. As a product leader and a solo founder, I live or die by how fast I can turn fuzzy ideas into crisp value propositions, roadmaps, and launch plans. That's why I love Miro's innovation workspace. It drops an AI co-pilot inside the canvas, so stickies, screenshots, and brainstorm bullets can become usable diagrams, product briefs, and even prototypes in minutes.

19:55Your team can dive in, riff, and iterate. And because the board feels like a digital playground, everyone has fun while you cut cycle time by a third. Miro lets humans and AI play to their strengths so that great ideas ship faster and happier. Help your teams get great done with Miro. Check out Miro.com to find out how. That's M-I-R-O.com. So while you're pulling that up, I have to ask, did you try LennyBot yet? Because I know he has his own bot with all of these newsletters and some content loaded in, but you wanted something really precise, right? So you created your own?

20:37Amir Klein:I had something really specific because I know what it is that I needed. So I took in her newsletters because he hosted her on his podcast. She also referenced a bunch of books. I looked for those books online and put them in here. I never read them, but I just put them in here. I was like, okay, you must know then how writing and delivering is. And then you have this here, this instructions that like this GPT rewrites Slack messages to be more concise, to be more clear, to be readable, maintains the natural voice of the user because I want anyone in my company, anyone in general to use this GPT so it doesn't make it super AI.

21:14Amir Klein:Like I said here as well, like, you know, avoid too much dashes. Don't give so much bullet points, all that stuff like that. And then I come into here so often, I'll write a message out on Slack to like myself. I'll copy it, paste it in here and it does a really good job of making it super concise, sticking to these guidelines of like delivering to the point, having a really good follow-up and lead-up as to what the next section will be. It is a lifesaver. People respond to my Slack messages more now. So I know that you didn't read the books, but I do have to recommend, as somebody with a liberal arts degree, those are exceptional books on writing.

21:48So I highly recommend reading them. We'll link to them and the newsletter and Wes's website in the show notes. But what I love about the approach here is a lot of times people get sort of professional development feedback like you got. Your writing's great. it just needs to be shorter. Like you just write too much. And a lot of people don't really build for themselves a way to coach into improving those skills, right? They're like, thanks for the feedback, but I'm not going to sit with my boss for an hour and have them like edit my Slack messages and edit my PRDs. Maybe I'll read a book or maybe I can do some research, but it's a hard thing to put into practice into your day to day.

22:30And so what I like about this is you took some coaching feedback that you got. And you're like, cool, I'm going to create myself a coach for this very specific development piece that I need to work on. And I'm going to rely on experts that I think are great. And then I'm going to use it day to day to actually make things better. And then what I like to, again, about what you said is, and then I share it with my colleagues, because I'm sure I'm not the only one that has this problem. So I think that's a really, Yeah, really nice process.