Welcome! If you're new to this blog, my name is Dani. I am a product manager with more than 20 years of experience building digital products. In this series, I will share my journey building a product solo (it's just me) from zero to release. I am technically literate, but I am not a developer, and I cannot write code. I am also not a designer, and I don't know any clever Figma shortcuts. Instead, in this series, I'll be mainly using modern AI tools that are based on natural language and available to anyone for a moderate fee. I have no affiliation with any of the tools, and no one (sadly) is paying me for this series. Quite the opposite: it is me who pays for the tools and (not infrequent) mistakes I make while using them.
Why build?
Is the question! Building software is time-consuming, (often) expensive, (at times) frustrating, (seldom) immediately rewarding and always bad for your eyesight. Why do it then? Here are (for me) reasons to and NOT to build software.Don't build software:
- To get rich
- To get famous
- To impress friends
- To save the world
- Learn
- Test ideas
- Solve problems
- Have fun
A solo builder can create digital products from 0 to market launch and break even on costs.
Timeframe: 6 months. Budget: $200 dollars a month and no more than 30 hours of work a week.
Ideas
For years, I've been advocating that ideas are cheap and implementation is what matters most. Vibe coding tools change this long-standing equation meaningfully. Writing code got much cheaper recently with tools like ChatGPT Codex, Claude Code and Google Antigravity. Using those tools, one can test most of their ideas quickly and cheaply. But the ideas also didn't get more expensive. Deciding on the ideas to commit to is where the costs have risen substantially. Those are mainly cognitive and motivational costs.If you chat with AI about your ideas, they might make it sound like the best thing ever. If you dig deeper, you almost certainly discover already existing endless competition. And if you don't, if the niche looks empty, it most often means the idea is not worth building.
So how do you decide? Two ways:
You scrupulously analyse all available data, make extensive product discovery, build in circular feedback loops and pivot based on insights
or
You go on vibes. Literally, you commit to the idea that feels good.
And I choose that second option. Which was a big surprise to me, as if you read this blog, you won't find a bigger proponent of product discovery, data-informed building and feedback loops. And yet, when it was my time and my money on the line, I chose to go on vibes. Mainly because it is only me. If I "fail", I would have "wasted" some time and money that I can afford to "waste". That might not be the case for you, so pick wisely, my friend.
Why did I pick a game?
Yes, for my first end-to-end vibe coding product, I've chosen to create a video game. Partly for personal reasons, my mum was playing a similar game that was riddled with ads that annoyed her. Partly, I picked a game because it fits to test my hypothesis.A game includes art, mechanics, UX. And I am not an expert in any of these, so I will have to rely on AI tools a lot.
Releasing a game means dealing with various stores and distribution systems.
A mobile game needs to support multiple devices, screens and platforms.
A paid game has a clear measure of success - sales- that could be compared to production costs.
Even though gaming is a very competitive market, no one really NEEDS to buy a game; people do that because they want to.
So, behold, my first game - Crazy Balls.
Crazy Balls is a board puzzle game about matching same-colour balls, opening space and chasing a better score with every run.
It was a good choice for the first game because:
- It's a well-known and established game mechanic that could be tested against existing popular titles
- Core gameplay is simple enough for the broadest audience, from young to oldest casual mobile gamers
- The scope of the game is small enough to fit within my timeframe
- A chance for success requires either visual or gameplay differentiation
- I just really liked the name
So that's how it started
When I settled on the idea, I had the following setup in terms of tools:ChatGPT Plus subscription (GPT 4.5 at the time), Codex (web version), Antigravity IDE (with a free Gemini and super low limits), no design tool, no macOS device to do iOS development.
Tune in to see my first steps and first mistakes.
Next in the series:
- Choosing a game engine and setting up tools
- First iterations and first setbacks
- Functional build and playtesting mechanics
- Custom assets and screen variety pains
- Preparations for publishing and where the most time went
- App store at last and initial promotion
- Costs breakdown and major learnings
- Hypothesis validation and future plans