Rapid Prototyping Is The New Research

We have a little special occasion here. We're in person, it's a warm, sunny day, finally. And we're sitting in Alex's kitchen. Yeah, so you might hear some birdies chirp outside. The audio quality is maybe not the usual. I will try and give my best in post-production, but at least we're here. And we're catching up on a lot of things, right? So Alex, let's jump right into it. I mean, you're the one with all the updates, right? My life didn't really change. Didn't it? No, same, same, same. Well, I mean, to start catching up on the past, I'm working on a little side project. And to be honest, I was learning a lot about product management, and I didn't anticipate to learn that much about product, because my point of view on building my own product has generally, building products, but especially due to the fact that I'm building an own little side project, has changed a lot. So out of all the advice that we gave in this podcast so far? Yeah, forget it. Okay, yeah. Starting from scratch? Starting from scratch today. Shit. Okay. Sorry. So where do we want to start? I was recently exploring a lot about the AI topic, as well as automations in general. And I just stumbled upon a great feature that I got, a great tool that someone recommended me, which is make.com. Not sure if you've heard about it. No. I think you know Zapier, right? Yes. Yeah. So it's maybe a little bit more technical and more enhanced in its capabilities. And I just started building a little survey for people. So I didn't jump, so it's maybe not something I can share now with the audience. So let's say you are trying to figure out what you want to do with your life. You want to maybe change your job, or you want to figure out what kind of strengths you have in order to do something meaningful in life. Because many people are struggling with that, right? I mean, if you just look at our job market, people are constantly changing jobs. And we read many times about complaints, about mental health issues and workplace issues that are happening. So people are changing jobs. So I thought, okay, maybe I can develop a little tool that help people to figure out what they want and what to focus on. So I started experiencing with make.com. So I built a little survey, a little database, and started using chat GPT to evaluate data and give some meaningful insights about the personality traits and strengths of people. So with that said, I had a lot of fun and I learned a lot about building such a product. And then I just asked myself, okay, how can I start making people use that? So just like for beginners, right? Because like my AI and I mean, we talked a bit about AI in our previous episodes. We need to catch up on that as well. I wouldn't say I'm ignorant about it. I even spoke at a conference about AI. So you promoted something that you haven't tried out yet. The thing is, so my AI knowledge stops very shortly after chat GPT and mid-journey, right? So I understand generative AI, more like language-based stuff, right? I mean, again, I usually tend to use it more as like a second opinion, an advisor, something to get me started on a project, helping me with brainstorming, getting some first ideas out. But that's pretty much where I am. So I mean, if you're talking about make.com, I'm thinking of mate, which is like this furniture shop that unfortunately went bankrupt. I think they still exist, but they got acquired. Great, great place. If you need furniture, they're like the European China. Listen to the designer. He knows everything. They have cool stuff, but a lot of it is copied from actual design pieces. So I shouldn't be advertising it. No, come on. Ikea has their own. Ikea is pretty iconic, right? You're right. Take pieces like we should make an episode about Ikea. Oh, that's a good point. Okay, let's add it to the backlog. Okay, yeah. Vote it up if you want it sooner rather than later. Otherwise, we might forget. No, but the point is like make.com, what does it do? Yeah, so it's an automation tool. And so what you can do is actually, from my opinion, you can integrate a lot of stuff. For example, you can use webhook triggers to start certain automations. For example, whenever someone submits a file from my survey, it will create the entry in my database and that triggers a webhook. So that means the make.com automation I've built will receive the data. So now talking about AI, I mean, there are many steps in between that are happening, but I don't want to make it too technically. But apart from the chat GPT interface that you are using, there's also this open AI API, which means you can communicate directly with the API. So what you can do is in the tools such as make.com, I mean, for sure, you can also build it yourself as an engineer, which I'm obviously not. So you can then send these informations to the API. So make.com itself isn't AI. That's just like automation. You're using AI through make.com. Exactly. Okay, gotcha. Okay, that helps me a lot. Yeah. So and then I'm going to send the data, obviously with my prompting and my magic behind it to the API and I will get results. And these results will be written back into my database with make.com. Okay. So that's what make.com is made for. Makes sense. And then I'm using a Google product to create a PDF result, which I have designed by myself. And this will be automatically sent to the person who filled out the survey. And the cool thing with that is, I mean, since I'm not only building products, but I also want to be profitable, you will get a free version that tells you some basics. And if you want to get the advanced analysis, you have to pay for that. And the cool thing is, and I know it sounds maybe very easy for people, but there is a payment link integrated. And once you buy it, it will trigger another automation and create for you the enhanced version. And what I just realized is because this is to me building products on steroids. Back then in 2014, when I started working at my first company, which was what was called ShopGate. So they were building mobile apps for people. We had an integration team that was just integrating payment methods, like Visa, MasterCard. I mean, everything needed to be integrated manually back then to make it available on mobile websites and apps. And now I can just go to Stripe, create my account, connect it with Make and use it to create payment links and add those payment links to my PDFs that I create for people. I mean, this is nuts. To me, it's like, I mean, back then we had an engineering team integrating that stuff. And I know times have changed, but I can't just stress enough how crazy it is what you can do these days without coding knowledge. But who's the winner in this, actually? Me, obviously. I mean, yes, you're one of them. But obviously, it only works if there is proper integrations, right? So I feel like Stripe writing the integration for Make, they are the ones winning, right? Because at the end, it's Stripe replacing your integration team to some extent, because they provide you with integration and they make it easier. Absolutely. I mean, I'm thinking of Zapier, right? Because I haven't used Make. But I'm thinking of Zapier as a comparison. So I mean, it's basically also their advancement or their integrations that then make a lot of this possible, right? I feel like there is a massive opportunity for companies to build these integrations to make sure that their services remain viable to the audience by allowing people to then use it through automation tools like Make and Zapier. And if you look at Zapier or Make, it doesn't matter. I mean, there are three ways how this integration is getting developed, either by the company itself or by the community or by the provider. Nevertheless, I mean, as you said, I mean, I as a person who wants to build an automation, a tool, a product, however you call it, can easily make use of it. And the good thing is, the more time passes, the more integrations will be available, obviously. So, but, sorry, just like me to, well, I shouldn't use but, that's what some wise people told me. And, and I shouldn't say, give me your but, no, but there you go again. There is still like, there is more people in the value chain, right? Like if, let's say back in the old days when there wasn't a lot of open integrations and a lot of APIs that you could just like directly tie into it, either through having a team that implements those or for using a service like this, you would potentially write your own payment system, right? And then like, it's one third party less. So it's one cost factor less, it's one revenue share less, or one percentage point somewhere, right? So pretty much the value chain itself loses a bit. Yeah. I mean, depends on what you define as a value chain. More people make money with it. Yeah. And you make less. Yeah. Okay. Wait, wait, wait, wait, wait, wait. I don't make less. Yeah. Because otherwise you wouldn't have made it. Right. So it's like. Sorry. I mean, you're talking about me as the user? Yeah. I mean, I have a direct connection to Stripe, which means I pay Stripe. Yeah. But you pay Stripe. Exactly. You don't have your own payment service. No. You would make more if you would. Okay. But here's the thing. I mean, but even these days, people are. I'm just trying to understand the ecosystem, right? Because it just shows that for me, if I have a company now and I want to stay relevant in the world, I would make sure that I have these integrations so that more people would use me for whatever I started doing. So yeah, that I just keep my relevance, right? And then there is a ton of people who can use my product. But you should bear in mind, I mean, Stripe SDKs or PayPal SDKs have been out there for a long time. Right. But now they're accessible for people who cannot code or more accessible than they used to be. And so here's the thing. To just catch up on the whole topic of building products. So one of my big learnings is there are many things that I can build myself. And the fun fact is why I was developing that feature and just came across multiple problems. So for example, I have a website and I want to track how people are doing the survey and how they are visiting the website. So I obviously use Google Analytics. Have you ever checked out Google Analytics, Alex? Yeah. And how do you like it? From a design perspective and user experience perspective? Yeah, I feel the same way. It's terrible, right? My biggest problem is more with the company is not integrated and that they usually have difficulties trusting even the accuracy of it. Right. But that's. So that's. Yeah. And the thing is, I was pissed about it because it's hard to understand. I don't know where to click and what to do, but I just came across the idea, hey, I could just download the CSV file. And then I can just create my little make automation where I upload it, give it to chat GPT and tell chat GPT, OK, I want to have this data. Please give me a little report. And chat GPT does the work for me and explains to me, explains it to me in a way that I can easily understand it. I mean, for sure, the accuracy is always a problem. No. Yeah. However. But at least it's more. It's more user friendly. Right. Yeah. And the cool thing about that is, I mean, you can make a sass out of it right away if you like. Yeah. So and that's the cool thing. So I just realized, OK, with all the problems that I'm facing during that process, there are many other problems that I can just easily solve with new automations. So and this is already like, wow, that makes so much fun to to explore. Same thing for Facebook ads. I mean, this is the next topic I wanted to talk about. If you do Facebook ads and if you have the Facebook ad manager, which is the worst, worst product I've ever seen designed by. I don't know who designed that. I mean, how much money does a fucking sorry, a designer make at Facebook, at Meta? And if you look at this fucking shitty user experience, I think I can kill yourself. I can probably hire ten designers for the price of one person. Hell. Fang. So same experience there. Same solution there. Yeah. So that's the product life. I think many people can relate to that. I'm not sure if that's where you want to take the conversation, but I'm just like in. We always end up in a rant, right? No, I'm. Yeah. Yes. Which which I like. But I'm in this. I'm trying to understand a little bit more on what all of this means also for the industry. Right. Because like what you're telling me also is that, OK, I can now make my automations to solve some of the issues that other like the companies like Facebook and Google didn't solve when they built a product. So there is like the side of why can't Facebook not just make their product better? Why can't they not use natural language to explain whatever you're seeing or whatever you're doing? Or why can't Google not do this? I mean, Google did for a period of time, I think a couple of years ago, they already had like this AI interpreter where you would just like type in a question and they give you the answer, which was kind of nice. But at the end, like we're building workaround, like we're leveraging AI to build workaround to solve stuff that ideally a product company would have solved. Exactly. Exactly. So what do we learn from this? What do we learn from this? I mean, again, it goes back to like, OK, who benefits from it? Right. Who benefits? Obviously, how many products out there are good, are bad, are failing, are succeeding? Yeah. Yeah. I mean, the thing is, if you find a use case that solves a problem that people have and I mean, many people are using Facebook ads, many people are using Google Analytics. No, of course. Yeah. And if you find a way to make this user experience for them better, I mean, of course, you're putting a layer on top of the actual problem, which is in, for example, the Google Analytics interface. Nevertheless, due to the fact that, for example, Facebook or Google are not fixing that immediately, more actually for decades. Yeah. I mean, workarounds will always appear. But so people like what we now see is there is more or there is an additional revenue stream evolving for people that sell workarounds to make users life easier. But here's the thing. Because nobody is building like AI isn't necessarily building a new analytics tool like you're not using AI to build analytics tool. You actually create a new interface that helps less technical people work with the tools that are available. Right. So just an FYI, I am not building a SaaS regarding this analytics, it's just for myself. No. Yeah. But I mean, it's pretty cool. It is. It is. And but the thing that I was actually was planning to go to today is I just realized that the way I was thinking about user research and approaching user research has changed. And that's something that I wanted to talk to you about. And I'm also happy to hear your opinion, because, I mean, back in the days. I came up with an idea, I did some user research, I maybe talked to people, I created some wireframes as a product manager. You know, I don't want to say I designed or some low fidelity, yeah, a low fidelity product. I mean, I believe I'm a good designer. You can, you can. But nevertheless, I mean, there are certain steps, right? So talking to people, showing them something, trying it out, getting feedback, etc., etc. So these days I have the feeling, I mean, why are we talking? I could already build a little interface myself and just throw it on Facebook with some Facebook ads and get feedback. Which means something that I've explored is that the way or the possibilities that I'm having these days to build, measure, learn something is drastically decreasing. In this workflow, help me understand, right, like, how do you build the interface? Let's talk about a survey. Yeah. So, I mean, a survey is something that I've created quickly in type form or something like that. So I put, take the HTML, put it on my website. Which means I have a publicly accessible place where you can try out this product, the service, however you call it. So that's my interface. And then I can just throw it on Facebook ads. I can post it on my LinkedIn personal account, doesn't matter, right? So people start interacting with it and I'm going to collect data. I mean, of course, there are different types of data, qualitative versus quantitative. But I'm quickly achieving something with maybe within 24 hours. While back in the days, I would have maybe spent more thoughts on, okay, let's think about the survey. And of course, I mean, the survey, I'm not telling that the first version of the survey is perfect, but I will figure it out once it hits the fan, right? I mean, at the end, it's rapid prototyping, but just like a different approach to it, right? Good that you say that. That was the word I was missing. Yeah, it's rapid prototyping in a way. I mean, at the end, things like... oh, I don't know, there's a ton of products out there that use Typeform as a central part of their product. Or Notion. Yeah, because it's just like easy to get something out of the door. And I mean, you can speed it up further, right? By saying, okay, I'm, and that's where my general or the way I look at AI comes in is like, it helps you just like, not starting from a blank sheet of paper, right? Yeah. So you throw in your idea, you see what comes back, and then you have a starting point to iterate on. Right. So instead of like having to go through, because like, even writing a survey takes a shit ton of time, if you want to improve it, right? So we didn't talk about that at all, you're right. I can just, but I can get first thoughts, right? I can leverage the AI to help me like constructing it, I have something testable. But not only that, I mean, you can also make sure that it fulfills certain standards, right? I mean, statistically significant data. I mean, there are many approaches how to build a survey, which I'm still learning. But so there are many ways how to create a survey. And of course, the AI helps you. But we need to distinguish between the AI that helps you, for example, prototyping and also doing rapid user research in a way, right? Versus the automation tools that help you getting it on track. However, the question I'm still asking myself, I mean, of course, building like a huge product, like a search engine, like Google or something. I don't think you will easily achieve this now with an automation tool like Make. So I'm talking about a small product level. As of now, I don't know which direction it will develop. But what I see is that it's scalable. Because the thing is, like, you're not creating the tools that you're currently using to create this, right? You're not rapid prototyping a new type form. You're not rapid prototyping a new analytics tool. You're not rapid prototyping a new notion. I mean, rapid prototyping a survey tool is something that is possible. That's where I'm thinking about it while we're talking. I could develop my own prototyping tool. Yeah. I mean, you definitely could. I know there's a ton of people like with the right tools, you can build interfaces and you can launch an app. I mean, if you have like a good front-end builder tool. I mean, I'm not talking about Squarespace, obviously. But yeah, I mean, things are getting more and more possible. I know our CPTO. What a title. Yeah, I am. Untitled. No, but I mean, our CPTO, just for an experiment and also a little bit like to challenge the company. He built the first version of our mobile app before we started developing it with AI. I love it. To prove the point of how quick you can release an app. And obviously, like, he didn't tie it into our whole database and so on, right? But it was like a week project of, hey, here you actually have a usable app that runs on my phone. Right. And I love what you're saying, because I think, I'm not sure which book it was. I think The Hard Things About Hard Things. Or what was it again called? I have it outside. Okay. I just started reading it. Okay, good. Because there's like one chapter, sorry to spoil it, as far as I remember. Maybe it was a different book, but it doesn't matter. Where the author said, if you want to stay in business, grab a little bit of money, hire a team of four to eight people and let them build a competitive product within your company. Yeah. And that's exactly the use case, right? So just sit down. I was on such a team. I love that so much. Ten years ago when I worked for ShakeStick. I would do that anytime. Just if you want me to do. We had a team of eight people versus an IT organization of 300. Rebuild the product from scratch. And you know what the craziest thing is? So let's take a product like SAP. We both worked there, right? So if they provide us with a webhook to the database, I mean, then everything is possible. I mean, if we have the webhook, if we get the data from the back end, we can build everything by ourselves. But now the question I have is like. And I think that's just interesting. Beyond the rapid prototyping point. Yeah. As of today, an AI can't replace the skills of an interface designer. Yeah, absolutely not. It can't. It also can't really replace creativity. It can enhance certain skills from people like me and you, right? I would say every designer should use a combination of working with AI to get, even just to have an additional sparing partner. Especially if you're alone as a designer, right? Like you can bounce ideas off. It's a really cool tool. But I don't trust an AI to come up with something innovative. And then an AI being able to then reproduce it in a way that you have consistency and that you can actually develop a product. Agree, Alex. Agree. I think as of now, the AI is a tool that leverage your skill set if you use it correctly. And the thing is also, I mean, even these days, which I think is also very interesting, maybe my brain works a little bit different. Most important thing is to know how to use the AI and how to prompt. I mean, obviously, prompt engineering. There are many articles about it. I think I haven't found a real good article yet where I would say, hey, this is something that I would recommend to you. But if you know how to talk to the AI and if you understand how the AI works, you can leverage that tool even more. Yeah. And I think more importantly, you need to have an understanding. Yeah, absolutely. You cannot expect that the AI knows more than you do. Not yet. Because, yeah, not yet. And this might change. But I mean, in order to properly prompt, you need to know what you want. Exactly. You need to have an understanding of the process. You need to know how to give. I mean, at the end, it comes down to also just like applying the processes you would use as human, but leveraging the artificial intelligence to do it. Right. But the process is the same. And if you don't know the process, it's hard for the machine to figure it out. It's same like leading people. Right. If the designer comes to you and you should tell this designer, literally build me an interface that can mean everything and nothing for this person. Right. And the same is with the AI. So you need to exactly tell the AI what you want. Exactly. Set a vision, set a strategy. And now we can refer to other episodes. Nice. So to sum it up. So my key takeaway was I can build products myself in a very quick and rapid way. During that process, I discovered many more problems, which I realized I can fix also by myself with automation tools and AI and exactly in that order. And I'm feeling like I am. So if I would tell you, hey, Alex, you can go to the gym tomorrow and you can do one thousand kilo bench press. That's the same how I feel right now when it comes to product development, because I can be fast, I can be very innovative as well. I can just try out and I can fail fast. I can make mistakes, I can try things out and I'm learning and learning and learning. And I think that's just something that I remembered during the last couple of months that product management is not about knowing, it's about learning. I mean, already that's kind of amazing or a quotable sentence, let's put it like that, because I think a lot of people forgot about this. It humbled me, honestly. Yeah, but I mean, it's all about like testing things out quickly. I think it's good for people to know that you can prototype by using a ton of like different services and just like tying them together to get a first feeling of a product, to get a first feeling of a market fit. Again, rapid prototype instead of like doing it on your own. And I think the one thing that I also liked, which maybe wasn't like the main purpose of what you were saying, but I feel like it's also an easy way to fix certain inefficiencies with tools that haven't been solved yet, because you just like integrate your tool and you build whatever around it that works for you to get the results that you need. You don't have to accept shitty products anymore. I mean, I do feel like shitty products shouldn't exist because people should build the right products. But that's probably something for another episode. Exactly. So and if you made it until now, don't forget to hit the like and subscribe button. Yes. Talk to you soon. Bye bye. This was the Product Bakery. All links can be found in the podcast description and make sure to follow and subscribe for weekly episodes on all podcast platforms as well as YouTube.