Vibe Coder is NOT my identity, but obviously I like trying the tools.
Here are some interesting demos using mostly AI-generated code, and in some cases, they’ve grown beyond simple demos.
- Gemini for Android Studio
- Google Antigravity
- Various LLMs, especially "Claude Opus 4.5" by Anthropic and "Gemini" by Google
- Claude Code
- and more
Find this app on Google Play Store
I've built multiple native Android apps, and published one of them on the Play Store: Plus Coords GPS Data
It's a very basic offline application that shows GPS data, coordinates and that "plus code" (open location).
If you’re in a place where your phone can send text messages but doesn’t have Internet access, you can even generate a link offline and send that link over a text message. This is also true if your data allowance is too low for your usage.
See also the remainder of this page for a PWA app that you can install on a phone right from the website.
Repository: daviddallet/antigravity-tools-linux
"Tools for those using Google Antigravity on Linux."
Especially one Bash script and one Python script related to model quota checking. Scripts were mostly made with the help of Antigravity itself.
Status: In Development – Access available upon request
Github Repository: pwasuite/youtubeplaylistexporter
Tired of the infinite scroll when managing large YouTube playlists? This tool provides a more efficient alternative to Google Takeout, allowing you to seamlessly export your playlist data to CSV or JSON.
A public version is coming soon, Google must validate things first due to usage of their API. If you are reviewing my portfolio and would like early access, please contact me for a demo link. If you ask for a demo link: please provide me email addresses associated to reviewers. They must have a youtube channel associated to that email. They don't need to have published anything.
If you are confortable with cloning a github repo, using node, running things on localhost, and asking a token in Google Cloud Console, you may also try the app locally.
Test it online: pdf.pwasuite.com
Github Repository: pwasuite/pdftools
Provides many tools to manipulate PDF files in the browser: compress PDF (minify or reduce size), merge multiple PDF documents into one, split PDF into pages, extract a range of pages, convert to grayscale, change page format.
This project was not started from scratch, I forked an existing open source project using Ghostscript WASM in the browser and then I used Antigravity-assisted development to completely change the UI and add new features.
Before the fork, the page had a single function: PDF compression.
I also transformed the demo into an installable Progressive Web App. You can use the website, installation is optional.
If you have seen "similar" projects, what’s less common is that the files are never uploaded to a server — everything runs locally in your browser. This is a key feature providing the highest level of privacy.
Test it online: fvibe1.netlify.app
Just a demo website showing some Formula 1 results in different ways. It also plays national anthems for drivers and constructors. Use various APIs.
Test it online: namedetective.surge.sh
As AI describes it: "A detective-themed name profiler that calls 4 public APIs in parallel (Agify, Genderize, Nationalize, REST Countries) and displays predictions with animated confidence bars. Click "EXPLORE CASE FILE" to reveal Postman-style HTTP details: endpoints, status codes, response times, and raw JSON. Dark glassmorphism UI with a playful "CLASSIFIED" aesthetic. For demo purposes only — don't take predictions seriously!"
Test it online: ghprofilecard.netlify.app
As an LLM describes it: A sleek profile card generator that fetches GitHub user data and displays it in a beautiful, shareable format. Enter any GitHub username to see their stats (repos, followers, stars), top repositories with star counts, and a language breakdown chart. Features four color themes (Dark, Light, Ocean, Forest), animated stat counters, and PNG export functionality. Uses the GitHub REST API with smart caching and CORS proxy fallbacks for reliability.
Test it online: DESKTOP VERSION ONLY - wikipulsetrade.netlify.app
Important: not suitable for mobile screens or for metered connections but you have plenty of other demos ;)
As an LLM describes it: Watch Wikipedia edits explode in real-time, then trade them like stocks! Every edit type becomes a tradeable ticker — buy low, sell high, and try to beat the market of human knowledge. Built with the Wikimedia EventStreams API.
Here is my LinkedIn, as you can see my carreer in software development started long before ChatGPT.
And as a software developer experienced in backend, I would strongly recommend you to resist to the urge of adding lots of lines of code if you cannot review them properly.
I believe humans are still essential in software development in 2026.
I do believe that you should not have a blind trust in AI.
However, it's definitely an area worth exploring and experimenting.