LTB #6: I built a tool in 49 minutes just by chatting with AI 🤯

TL;DR

I recently built a web app, momtest.ai, to help me with customer discovery using principles from “The Mom Test.” In 49 minutes, with no coding or design, I used AI tools to build, design, and deploy the app. This experience showed how far AI tools have come, making it possible for anyone to create a solution with just conversations and minimal technical knowledge. The takeaway? Test new AI tools, solve your own pain points, and potentially build something valuable—without needing a developer or designer.

p.s. There is an audio version if you press Listen Online in upper corner.

Here is what I am covering today:
1. Tell me more…
2. Why is this story important?
3. How did I do it?
4. Conclusion

Tell me more…

Recently, I Joined a company as a Lead Product Manager (which will be covered in the following newsletter) to shake up their product line. I’m currently working on several bold new products that could shift the industry, and to make sure we’re on the right track, I need feedback from potential customers. That’s why I’m attending a conference at the end of October and preparing the right questions to ask there.

Customer Discovery is a tricky process as we’re often biased, eager to sell, and hoping customers will love what we’ve built. To get honest feedback, we have to ask neutral, non-leading questions.

I started using the book The Mom Test to create my questions, but constantly tweaking and validating them through GPT became cumbersome. That’s when I decided to build a tool - momtest.ai - to make this process easier for me and maybe help someone else.

But that’s not the main story of this newsletter - the way I created it is.

Why is this story important?

The moment I decided I would create a Web app, I decided that I would shoot three rabbits with one shot:

1. I will learn how OpenAI API works and what you can do to manipulate it.

2. I will try our currently trendiest AI tools for web design and development - I will discuss them below.

3. Will get a tool for myself to work faster and better

And I am blown away by this experience.

Intentionally, I set up a timer to see how much time it would take to make this web app - from a white page to a fully deployed page on the server. I was ready to write some code, but only when I reached a dead end with AI tools (which I did not reach).

It took me 49 minutes to deploy the page - I did not make any designs or write a single line of code. Just chatting with Ai - that is mind-blowing…

I have heard promises that everyone can develop something using AI. However, all the tools I tried before were just bold promises - you still had to do code and design yourself a lot. Still, this experience showed that we are a few good AI tool integrations away from the ability to create tools and solutions by just having conversations. Like you would speak to developer, but there is no developer - just AI and you.

Now, everyone should spend time playing around with AI tools to understand how exactly they work and what can and cannot be done - test all new tools and see how they are better.

Why?

Yes, I lost 49 minutes of my time creating this tool, but now that I have it, I will be much more productive in the future as I have a tool I need, but there is no such tool out there. This is just one example - if you keep your eyes open, you will notice that there are still so many frictions in your daily life where you could solve them with the right tools, and the best part is that you do not need to wait anymore for someone to create a solution for you - you can do it your self. Yes, YOU.

If you make these tools available to the public, you will also open a window of opportunity for yourself - maybe something will take off, and many more people would love to solve the same friction for them.

There are countless ideas that have remained undeveloped simply because of excuses like "I don't know how to make it," "I can't find a developer to create a prototype," "I don't know how to design," "I don't have money," "I don't know where to find a dev" and so on. But after the experience I just had, I refuse to accept these excuses. The tools are here, waiting for you to take action and try them out for yourself.

How did I do it?

Prompt

First, to validate if AI can actually give good results, I used chat.openai.com, uploaded a book PDF, and asked him to create a prompt. Then, I found this article on Google. Combined with the GPT prompt, the article, and my ideas, a few iterations later, I had a good version of the prompt that would actually give quite a good result. Took around 10 minutes.

Design

Now that I knew it could work and I had a prompt, I needed a design. I knew it would be simple, so I just went to v0.dev and prompted it for a design using the free plan. With the first prompt, the design was already OK, but I did two more prompts to tweak it a bit, and it was ready in 8 minutes.

And the design code comes with perfectly working code with all the hovers and animations.

Code

There is a lot of hype around Cursor.com, so I decided to try it. So, I opened a new project and created a blank Next.JS project. Then, I copied a simple line of code that v0 gives to the terminal.

Voila! I have a working website.

With the help of the AI assistant in this code editor, I effortlessly made the button work and connected it to OpenAI. All it took was pressing “Apply” to all the code changes it suggested, including some additional steps like inputting the OpenAI key.

If I had errors, I just pasted them into the chat. I asked AI to implement my Prompt correctly. There were some other minor tweaks, and it was ready - around 25 minutes.

Deploy

The deployment to the server took 6 minutes. I created a GitHub repository, pushed to it, and pasted the link into the Vercel app, and everything else happened automatically.

The website is done and available to the public.

Some prerequisites help me do it faster - knowing some Terminal commands, a bit of Next.js, and a bit of Git. But you can learn these things in 2-3 hours or just ask AI what to do.

Costs

v0.dev - free plan
Cursor.com has a free trial but costs 20 USD/m after the trial.
Vercel.app - free plan. I already have three websites on the free plan, which is enough for some time.
Domain name - I took 70USD/y, but I would not do it again - I would go for the .com domain, which is around 20USD/y
OpenAI credits - I added 20 USD but have spent 16 cents this far.

Total - 90 USD / y

The problem with this product is that I do not see how to monetize it, but I will keep it just for myself, but if there are other tools I will be building will always think about if there is a way to monetize them and 90 USD even if I paid for all the services, is so cheap.

Conclusion

I was ignoring all the hype, thinking - “it is just another hype,” but what I realized is that I cannot ignore it - I have to test all the new hype things, not because all of them are useful and great, but by trying to understand what these new things try to solve and how they are solving it, gives me a lot of ideas on how to shape products I am working on.

And, of course, I love building something - I just got superpowers from this experiment. Ten years ago, to prototype or test something, I had to find a Designer and Developer, now I can do it myself and if validated, only then build a team.

This is a massive change in how products and startups are made - the pace is going to become much faster, but entry into the game is also much lower. Before, it was about forming a team that makes MVP and then testing/developing the product, now, it will be about the founder creating a prototype using AI, and if it is validated, then team forming will happen.

So, next time you feel friction, try solving it for yourself.