TL;DR
If you are looking to build software and you don’t have any coding experience, go with Lovable.dev. It is at a point where anyone can quickly build no-code/low-code with few issues and minimal cost. That said, as complexity increases, if you don’t make intentional decisions about your architecture and database structure, things can turn into a spiderweb of complexity that becomes unmanageable. For these reasons, I would suggest non-technical founders skip learning how to code. Instead, focus on learning system architecture, database design, a conceptual understanding of how software works, and how UI component libraries function.
Sipping the AI Kool-Aid
“It’s like a boulder rolling down a hill—you can watch it and talk about it and scream and say shit, but you can’t stop it. It’s just a question of where it’s going to go.”
—Ken Kesey, quoted in The Electric Kool-Aid Acid Test by Tom Wolfe (1968)
Around the end of July, I decided to start experimenting with vibe coding. I have done some programming for fun and random tinkering here and there, but I haven’t programmed professionally for over 8 years since switching to product management. I wanted to see, with my background, what I could still build with the tools available and, if so, how much complexity it could handle before the system broke.
I started by researching options, but Lovable.dev quickly became the obvious answer. While not perfect, it has by far the easiest learning curve for non-technical users. It has many integrations that make complex functionality, such as authorization, manageable while enabling you to use platforms like Gmail, Twitch, and Patreon as authorizers. That is a single-case use, but with their recent integrations with Shopify, Atlassian, and Figma, it shows that they have, or are creating, the tools to let users build quickly.
In the five months I have been tinkering, I haven’t used any other tool as much as Lovable for building software products. Given how far ahead of the competition it is at the moment and how they keep adding on, I believe this trend will continue. I will eventually do a deeper dive into others, but right now, Lovable is my tool of choice and the one I recommend to others, as it continues to get better and better.
First Prototype
Required Context:
As this project was more about learning than profit, I wanted to keep it contained, so I created a tool for the popular video game series Football Manager (FM). The player manages a professional soccer team, controlling things like roster, training schedule, tactics, strategy, and more.
When a product requires too much context before you explain the name…
The FM community jokingly calls their in-game assistant manager their “AssMan”… so when I was searching for a domain, AssMan.ai seemed like the perfect domain name because it's both memorable and product-significant… So, yeah, this is why, rightfully so, my last boss removed my ability to name features.
Overview:
Prototype Goals:
With my prototype with Lovable, I had a few goals:
- Build something contained for MVP that:
- uses AI as a core component
- uses AI to handle an interactive conversation
- Try out PostHog
- Run my first Google Ads campaign
- Experiment with Google AdSense
Product Opportunity:
One of the best tools for user research is Reddit. In many subreddits, you will notice that users will ask a similar question or make a similar request over and over again. That is a signal that there may be an opportunity. In my case, I noticed people kept requesting feedback on their tactics. I, myself, have struggled with tactics in FM and previously used ChatGPT for feedback.
Executing:
Lovable built-in Integrations Used:
Lovable + Ionos = Simple domain purchasing and set up
Lovable + Supabase = Turnkey backend and DevOps, basic account management
Lovable + Gmail = Standard account creation (OAuth)
Manual Integrations Set Up:
OpenAI Platform = Analyzed images to produce feedback/suggestions, and an interactive chatbot
Misc Tools:
ChatGPT = Iterated constantly to assist with problem-solving, Created Logo, $20/m
PostHog = Amazing Product Tool for all your tracking needs, free
Google AdSense = I made $19.63 by showing ads on the website
Google Ads = I spent $1080.12… it doesn’t quite offset the AdSense revenue
Reddit Ads = I also spent $267.73… it furthers the offset…
GitHub = Code repository and branch management
If you use this link for Lovable, you will get 10 free credits added to your Lovable account, and earn me a small commission if you end up purchasing credits.
Results
I set out to:
- Build something contained for MVP that:
- uses AI as a core component - Done
- uses AI to handle an interactive conversation - Done
- Try out PostHog - Done, Love it
- Run my first Google Ads campaign - Done, a lot of good learnings
- Experiment with Google AdSense - Done, a lot of good learnings
All objectives achieved.
Analysis:
- 105,046 visitors to the website
- 125 users have created accounts
- 1,848 tactics analyzed
- 445 player snapshots reviewed
- 5,704 AI chats interactions
Cost:
Domain: $200 for 2 years; usually, you can get things for a lot cheaper, but .ai domains are more expensive right now.
Lovable: $400 is my rough estimate; I worked on a few things at once, so the exact cost isn’t clear.
ChatGPT: $44.21, I felt pretty reasonable
Ads: $1347.85
- Google Ads = $1080.12, 1.5 months
- Reddit Ads = $267.73, 1.5 months
Supabase: $51.26, I forgot the importance of paginating, which caused some extra fees before I realized and fixed it
PostHog: $27.62, should have been free, but didn’t set a proper limit for recordings
Total: $2,070.94
Profit?:
This project was always for learning first, and profit second. The exposure to cutting-edge software and experience I gained during this project were well worth the cost. This project gave me the confidence to see what can be created with the tools out there, take the leap of faith and leave my job, and start up some actual projects that I will discuss more in the future.
Closing Thoughts
This project showed me what is possible when you build solely through prompting. It is not a silver bullet, but it does move the line for when you genuinely need developers. A lot of what once required an developer can now be built, tested, and validated by anyone who knows how to use AI and prompting correctly.
It is extremely easy to build simple things or personal tools, but the moment you create something for customers, expectations change. Without real intention behind system design, architecture, and structure, complexity starts to make things fall apart.
If I were a non-technical founder trying to build software with AI, I would skip learning how to code. I would focus on system architecture, database design, a conceptual understanding of how code and data flow through a system, and how UI component libraries function. That is the foundation you need to use these tools intentionally.
The goal is not to become a developer. The goal is to get the product to a point where a developer can take over while you focus on customer value and product growth.
For transparency, I built this entire project in Lovable without any developer help. I was comfortable shipping because I avoided sensitive data and monetized through ads. If I were collecting anything sensitive or taking payments directly, I would have had a developer review the code before fully releasing it. You can control risk in an alpha or beta, where users understand the constraints, but once the product is in the wild and expectations shift, the margin for error disappears.
Top comments (0)