What was the very first project you vibecoded with AI?
On Product Hunt, I can see many people launching their products using "vibe-coding tools" like @Lovable , @bolt.new , or@Replit
I reckon many people who created something with them are usually developers who didn't have enough time for building a side idea before, but with AI, they could make it happen.
I am not very technical (know some coding/programming basics), but without the help of a tutorial or ChatGPT, I would hardly build a whole project.
Question not only for developers (but also tech newbies):
What was THE FIRST THING YOU VIBECODED?
Feel free to share the link or the picture
What tool did you use?
What was the most difficult part?
Did you earn any money with that?
Here is mine:
– It was supposed to be a directory of Bluesky tools– The most difficult parts were to define something + It also rewrote good parts of the code, so it was kind of a mess for me.
– I haven't earned any money because I haven't published the project. (I abandoned it. :D)
Replies
We built our entire website in @V0.dev for Flyweel, would love any feedback.
Maintenance and updates without burning through credits became tricky. Have started using @Cursor more and love the new release of @Dyad!
minimalist phone: creating folders
@flyweel_turner What will Flyweel be capable of? Track your finances? I would say that this is a more mobile app solution.
My first "vibecoding" was actually upgrading my trading system.
I had a quantitative trading bot for A-shares (Chinese stock market) that I'd built the traditional way. I wanted to refactor it from procedural code to a state machine architecture — cleaner logic for handling market conditions.
But back then it was the ancient era of AI coding: copy code → paste into ChatGPT → copy the response → paste back into VS Code → run it → hit a bug → copy the error → paste it back into ChatGPT → repeat. Ctrl+C, Ctrl+V was my most-used shortcut.
Then came Cursor, now Claude Code — and I grew up from a quant trader to an AI builder. I built two websites in one weekend, and now I'm learning SEO through vibecoding. Turns out "vibe-SEOing" is a thing too.
Visionnaire - an AI goal coaching app. You set a goal, and Arcus (my AI coach) builds you a real plan with strategy steps, a first action for today, and mini goals to track.
I used Claude as my main tool, with Node/Express on the backend, Supabase for the database and auth, and Groq to run the AI.
The hardest part was honestly not the code — it was figuring out how to get people to actually try something new. Distribution is brutal when you're starting from zero.
No money yet — it's free for now while I focus on getting real users and feedback. Just launched today on Product Hunt actually.
visionnaire.onrender.com if you want to try it.
My teenage son built a full SaaS on Bolt with Claude. A property zoning tool. Hardest part was the data layer, not the code. Every US city has different rules. Bolt crushed the scaffolding but domain expertise still mattered. Yes it makes money. Vibe coding eliminated the boilerplate, not the thinking.
Just vibe-coded a menu website! Made ordering a breeze with clean design. The AI handled most of the heavy lifting. Still tweaking, but loving the process!
I stumbled upon a website for ios development and started fooling around. I created a task list and a mood tracker using react native and expo to get a feeling (both for free on the app store), but then figured I should use swift for ios development.
I am using Claude a lot, it's easy to prototype and the results are always great. I have many ideas for small helpers and games but being a developer for decades I know that I won't finish anything in time using my own skills only. I just prepared the launch of my sudoku game, my fourth project as a vibe coding supported indie dev. It would not have been possible without AI support.
CCG — a GitHub Action that intercepts pull requests and uses AI to generate comprehension questions that block merging until the developer proves they understood their own code. Started as a frustration with rubber stamp reviews on AI generated PRs. Ended up filing a provisional patent on it. Repo: github.com/islandbytesio/commit_comprehension_gate
Tool: Claude. Most difficult part was prompt engineering the questions to be genuinely hard without being unfair — the model wanted to be too nice. No revenue yet but the patent is filed and a Pro tier is coming. Curious what others built.
The first product I "vibe coded" was a 100K line, DDD SaaS for PIFster, the Pay It Forward charity, which my wife and I founded and run. Donors donate as little as $1 dollar per month, then suggest and vote on charitable causes that are important to them, and the winning cause each month receives the community donation. I vibed it, but I do have a degree in CSE from 1989 and know enough coding to keep the AI within the guardrails. I also have a sweetheart deal on Gemini Pro, because of our nonprofit, so I use it as an external code reviewer and iteratively pass classes back and forth between Gemini and Windsurf (plus whatever LLM i'm using at the moment) to get the code into shape. The whole project is PHPStan L8 clean from line 1 and the frontend is fully typed using TypeScript and React 19. Build it right from the get-go was my motto. I also have maintained a dogmatic adherence to single responsibility classes, trying to keep most files under 300 lines or so. I say I vibed it, but I vibed it with standards. LOL. It has custom donations, voting, leaderboard totals, charity suggestions, referral rewards, gift cards, and more.
This won't count as vibe coding to everyone since I designed every component manually and just used AI to code.
Back in June 2025, I designed a mobile app called Spellnotes on @Figma and coded it on @bolt.new. It's a task management app inspired by character.ai. Instead of chatting with the characters, you input your task, such as "buy cheese for the BBQ" and the characters read your tasks back to you in their voices.
I only had very very limited coding knowledge before that so when I built it, I could barely understand why bugs were happening.
Did I make money off this? Hell no! I don't want to deal with what cursed things people are going to make AI characters say to them. This was also before I knew how much public backlash to consumer AI was out there.
I'm currently working on a non-AI misinformation platform Retrocodex and might make a non-AI version of Spellnotes where users can just upload photos of the character or object they want to see "delivering" their reminder without any AI modification. As useful as AI has been for software development, bringing AI personalities to the public is just too much trouble to deal with.