227. How to Use AI in Product Development: A Non-Techies Guide

If you have an idea for an app or a tech business, you can now make a test product in just 15 minutes using AI tools.

But, you have to know how to use them wisely, so you don’t make pretty products that nobody wants.

Listen to this episode to learn:

• The sweet spot for using AI in product development (from drafting specs to analysing user feedback)
• Which AI tools to use for prototyping right now
• A step-by-step process to validate your ideas faster using AI
• Common pitfalls to avoid (because AI won't save a half-baked idea!)

Timestamps

00:00 Introduction 

03:13 AI in Product Management

06:05 Creating Prototypes with AI

08:55 Testing Prototypes

12:08 Strategic Thinking Before Using AI

15:03 Practical Homework and Implementation

For more career & tech lessons, subscribe to Tech for Non-Techies on:

• Apple
• Spotify
• YouTube
• Amazon Podcasts
• Stitcher
• Pandora

FREE GUIDE: The Pragmatist’s Guide To Innovation (in business & in life)

by Sophia Matveeva, supported by researchers from The University of Chicago

Growth Through Innovation

If your organisation wants to drive revenue through innovation, book a call with us here.

Our workshops and innovation strategies have helped Constellation Brands, the Royal Bank of Canada and Oxford University.

Transcript

Sophia Matveeva (00:00.366)
If you have an idea for an app or a tech business, you can now make a test product in just 15 minutes using AI tools. But you have to know how to use them wisely so you don't make pretty products that nobody wants. And you'll learn how to do that in this episode.

Sophia Matveeva (00:22.382)
on Techies podcast. I'm your host, Tech entrepreneur, executive coach at Chicago Booth MBA, Safiya Matil. My aim here is to help you have a great career in the digital age. In a time when even your coffee shop has an app, you simply have to speak tech. On this podcast, I share core technology concepts, help you relate them to business outcomes, and most importantly,

share practical advice on what you can do to become a digital leader today. If you want to have a great career in the digital age, this podcast is for you. Hello, smart people. How are you today? You know, I'm feeling bittersweet. I've been teaching a Tech for Non-Technical Founders course for the last four weeks and we're in our last week and week five. And the day that this episode comes out, it's actually our last lesson and

In our last lesson, we're going to cover how to hire and work with a tech team. And this is quite an intense program and I really got to know my students over the past weeks and I'm just going to miss them. Hence this bittersweet feeling. And in this week's podcast lesson, I'm going to teach you some of the concepts that we've covered in the course so you can benefit too. And today we're covering how to use AI in product development.

One of my students in the course built her prototype in 15 minutes using AI tools. So if you've been working on an idea and want to test the market quickly and cheaply, then you absolutely need this lesson. So if you're a startup founder or a corporate innovator, keep listening. And if you're an investor and somebody comes to you with an untested idea, I just want you to know that they could have tested it with AI tools in 15 minutes. So the fact that they didn't, well,

It's not a good sign, is it? And before we continue, are you a subscriber to this show? If not, hit that subscribe button now to get excellent free lessons specifically for non-technical people about how to succeed, innovate and build products in the digital age. In this lesson, I'm going to tell you what you can do with AI tools right now to make your product better and your life easier without the usual hype.

Sophia Matveeva (02:44.546)
Because AI, it's great, but it can't be relied on for everything. And so let's begin with the status quo. How is product management done without AI tools? Well, I know this because I have done it in my previous company where our main product was an iPhone app. We actually spent hours on writing product specifications and user stories and release notes. And it was boring and it was tedious and it just had

to be done. But even though nobody likes writing product specification documents, I mean, even the title sounds boring, I've actually heard product leaders say, well, yes, there are all these AI tools around, but we can't trust it with our product decisions. And so basically then they just decide not to use AI at all. And that's throwing the baby out with the bathwater. So let's get clear about what AI is good at and what it isn't good at.

So here's an analogy that I like. Imagine that you're a chef and AI is your assistant who has read every recipe and every cookery book in the whole world and they've watched every cooking show ever made, but they have never actually cooked anything. So yes, their knowledge is useful and they could give you some tips, but you really can't rely on them to make a delicious meal. And some of their advice might be a bit insane.

But some of it might be absolutely brilliant. So when it comes to creating your app or your website, here's what I suggest you use AI for. Number one, drafting and formatting documentation like product specs. Number two, generating test cases. You can literally ask it to do that. Number three, analyzing user feedback. And we'll discuss that a bit later on in this episode. And number four, creating first drafts of user stories. And that's basically how people are going to use your product.

And now here are four things that you absolutely should not use AI for. So you should not use it to make final product decisions. The final decision is with you. Number two, do not use it to understand the specific dynamics of your market. Use it to help you, but the specific dynamics of your market is something that basically you really need to get.

Sophia Matveeva (05:07.374)
qualitative data for and quantitative data for. So you actually do need to have business instinct. Don't use it to predict user behavior unless you have fed the AI data because if it has data, then it can use past behavior to predict future behavior. But again, the past does not always create the future. Remember what happened in 2020? That was interesting. COVID and all that. Anyway, number four, do not use it to replace

user research. So I think lots of people think, now that I have AI, I do not have to speak to my customers. No, you actually still have to talk to your target customers and hear their feedback. So here's a bit more on my student who created her prototype in 15 minutes. So first she used a generative AI LLM. So basically chat GPT or Claude. And by the way, I prefer Claude because I think it writes better, but try both.

Anyway, so she used a generative AI LLM to help her describe her product succinctly and to create a summary of what she wants the product to do. Then she took that summary generated by Claude or by ChatGPT and then she put that into a generative AI tool specifically for prototyping. So there are literally now tools that make you a prototype. So the two most popular tools are called UI Wizard.

I think you're supposed to pronounce a wizard and honestly, it's a bit confusing. So I would rename it if I were them, but anyway, so it's kind of UI wizard and Galileo. You can literally write your prompt into them. tell it what you want the product to do, what it's going to be about. And then these tools will generate a bunch of screens for your app or your website. So basically the thing that exists in your head can begin existing as a clickable prototype. basically as a bunch of screens.

that you can start showing to people in just 15 minutes if you know how to do it right. So my student did this very quickly and very efficiently for various reasons. So number one, she understood exactly what she needed from each tool. And she also knew that she was not trying to create a perfect product. She was literally just creating something that she could put in front of users like tomorrow to start getting feedback.

Sophia Matveeva (07:32.308)
And this was literally unthinkable just 18 months ago. I remember when I had to work on making my first prototype, the actual creation of it, it took time. It took labor. It took specialist labor. It took working out how to use design software. And obviously there's a, these are barriers to entry, but now nobody has an excuse not to create a test product. So if you've been sitting on your sofa and thinking about a brilliant idea,

Now is the time to find out if it really is that brilliant. The proof of the pudding is in the eating. The only way you find out is by creating something and testing it with your target market. So now that this founder, she has created her prototype in 50 minutes. What's the plan now? So the plan now is not to give it to developers and to create it.

No, the plan now is to test it. You always have to test a prototype before you do anything else. So she has to give it to her target users and then interview them for their feedback. So in the not so olden days, you would have to interview people and record the interviews and then watch them again and make notes and try to spot patterns and user behavior. And based on your notes, based on what you've spotted,

You will then use that feedback to improve your product in the next version. I have done this. It's basically how the products in your phone have mostly been made. But now you can use AI to basically make this job easier and faster. So you still have to interview people. You still have to get feedback from your actual customers. But now here's what you can get AI to do. You can record your user interviews with permission, obviously.

and then use AI to transcribe and summarize key points. Then upload these transcripts into ChatGPT or Cloud and ask it to identify patterns and to give you product recommendations. But here's my word of warning to you. Do not take a half-baked idea and expect AI to give you a great product. The reason why this founder could create a prototype within 15 minutes is that she had spent months

Sophia Matveeva (09:49.6)
working on her idea and testing her market and getting customer feedback. And this is why she could tell the AI what to do so clearly and to get great results. So if you don't understand your problem and your user, AI will just help you build the wrong thing faster. And you don't want that. And I've seen this happen countless times with founders who jump straight into AI tools without doing their homework. So here is an example.

Imagine that you decide that you're going to make a productivity app and then you write the prompts in the LLM. And then after that, you take your prompt and you put it into UI Wizard or Galileo. And basically you'll get a pretty looking AI generated prototype for your productivity app. Will this be a successful product? What do you think? I will tell you, it will not be a successful product.

because it is too broad and it has not been thought through at all. And yes, there are lots of people who want to be more productive. I'm one of them. But in order to make a product that works, you have to be much more specific than, well, just make me something that will make people more productive. The way a surgeon is going to be more productive is different to how an economist is going to be more productive. Different segments have different problems and thus,

need different solutions. If you don't know whom you're solving the problem for, AI will not save you. So you still have to use your human brain. You have to do the strategic thinking first. So your brain first, AI second. So before you even think about opening chat GBT or Claude or UI wizard or Galileo or any other tool, here's what you need to understand. Number one.

your specific user segment and their actual behavior. So basically, what do your target customers actually do, not what do they say they do. Number two, the exact problem that you're solving and making something better is not specific enough. So if you just say, yes, we're making a better version of Uber. Well, how better? Why do people care? What is the exact problem that your product is going to solve? Number three.

Sophia Matveeva (12:08.384)
You need to know basic user experience principles like information hierarchy, user flows and decision architecture. And there are lots of episodes on design principles earlier on in this podcast. So just scroll back and listen to everything that has the word design in it or UX in it. Also listen to the earlier episodes on prototyping. So actually you understand what prototypes are. And then number four, you need to understand how to create and test your hypotheses about your users.

Only once you understand these things are you ready to use AI properly. And remember to always keep an eye on the difference between what's technically possible and what's actually useful. AI allows us to make product more quickly, but I implore you from the bottom of my heart, please do not make stuff just because you can. I could make a lasagna with a marzipan filling, but is that a good idea? No.

because it would be disgusting and nobody would want to eat it. So just because you can do it doesn't mean you should. And so now here is your homework. And my dear smart person, I do actually expect you to do this because you learn by doing, not just by listening to podcasts. Okay. So number one, take your most recent product problem. So is it that you have low user retention? Is it that...

People aren't signing up beyond a free trial. Is it that people aren't taking a specific action that you want them to take? So take whatever product problem you're experiencing. Write it down in plain English. Write it down badly in plain English. And then go to your AI tool. Go to Claude or ChatGPT. As I said, I prefer Claude, but honestly, try both. Some people prefer ChatGPT. So ask your AI to break that.

problem down into smaller components and then ask it to generate five different approaches to solving that problem. And then ask it to create test cases for each approach. So literally take your problem, tell the AI tool about it, tell it to break that problem down into smaller components and give you five different approaches to solving it. And then to give you test cases for each approach. And then you don't actually have to do

Sophia Matveeva (14:33.054)
any of what the AI says. You don't have to do it. These are suggestions. And you know what I like doing? I like thinking. I like thinking about what the AI gives me because sometimes it gives me something I hadn't thought of, or maybe, you know, it's 90 % obvious, but 10 % surprising. And then I focus on that 10 % and I extrapolate it. And then that's what I go with. So basically once you get the answer from this AI, if you've got a team,

Use that as a starting point for team discussions. If you don't have a team, then think about it yourself and basically choose the easiest and cheapest option and go with that. And I think you're going to find this interesting and you don't actually have to pay for any of these tools to get started. So use the free tier of Claw.Chat GPT. And honestly, the free tiers are good enough to be super useful. And this is basically the end of today's lesson.

Now it's up to you to implement it. And it's honestly easier than you think. So if you've never done any of this and you think this sounds hard, or I don't know, I'll do it tomorrow. Honestly, once you start saving large chunks of time, you will get addicted to these tools. And that's what's happened to me. And then you can use that time to do something much more pleasant than writing product specification documents. And as we wrap up, I have a question for you, my GSMart listener.

Have you found this lesson useful? I'm assuming you did because you are still listening. So if you did, can you please leave the show a rating and a review? Honestly, it really, really does help me because it takes me time and effort to create high quality learning content for you. So if you're benefiting from this work, then let's spread some karma. So please help me by leaving the show a rating and a review because it helps other people discover this work.

So thank you very much in advance. Have a wonderful day and I shall be back in your delightful SmartEars next week. Ciao!