My AI Lab Experiments 🤖
Welcome to the AI Lab, my digital playground for exploring the intersection of development and AI! We’re in an era where AI is becoming a powerful co-pilot for coders, and this is where I put it to the test. I’m using Gemini Canvas Pro as a creative partner to see just how quickly an idea can become a functional HTML application, aiming to smash through creative blocks and accelerate the entire workflow.
Below, you’ll find a growing collection of simple apps, each bootstrapped with AI assistance. Think of them as tangible proof-of-concepts that show how AI can handle brainstorming, generate boilerplate code, and speed up development. The goal is to share fun, practical examples of what’s possible. So, dive in, explore the projects, and get inspired to integrate an AI partner into your own tech stack!
Red Hat Documentation PDF extractor
Built with: Gemini Pro Canvas
I wanted to extract all PDFs from the Red Hat documentation for a specific product and version to use them as sources to create a NotebookLM. I thought that there should be a better way than doing it manually, link by link and then downloading the PDF.
I felt it was a good challenge to try Gemini Canvas coding for this purpose 🙂 and it actually worked, after some back-and-forth iterations it seems to be working fine.
You give the tool the main URL of the product documentation where you have the product and version (ex: https://docs.redhat.com/en/documentation/red_hat_openstack_services_on_openshift/18.0/ ) and it’ll scrap the main page for other URLs and get back all the pdf version URLs for you
MP3 Cutter App
Built with: Gemini Pro Canvas
I wanted to share another fun project I’ve been working on: a fully functional MP3 Cutter and Trimmer web application. You can use it to upload an MP3 file or load one from a direct URL, and it will display the audio waveform. From there, you can easily select a specific part of the audio, preview it, and then download just the cropped section as a new, high-quality MP3 file.
This application was created as an experiment to see what’s possible when collaborating on code with Google’s Gemini. My goal was to move beyond simple demos and build a genuinely useful tool that anyone can use directly in their browser, with no software to install.
Animated Maze Generator
Built with: Gemini Pro Canvas
Here’s another fascinating project I’ve been developing: an interactive web app that visualizes the creation of a perfect maze. Using the Recursive Backtracking algorithm, it carves out the maze’s paths in real-time, starting from a blank slate. I added a playful goal where a monkey needs to find its banana, and once the maze is complete, you can even reveal the optimal solution path.
This application was an experiment to explore how a complex, visual algorithm could be brought to life through collaboration with Google’s Gemini. My aim was to create an engaging tool that’s not just a puzzle, but also an educational demonstration of how intricate structures can emerge from simple rules. It’s a fun example of building a dynamic and polished app directly in the browser.
The logic behind the Recursive Backtracking algorithm is surprisingly simple and mirrors how you might explore a maze yourself. It starts at a random point and pushes forward in a random direction, carving a path until it can’t go any further. When it hits a dead end, it “backtracks” along its path until it finds a new, unvisited direction to explore. This process repeats until every cell has been visited, guaranteeing a perfect maze with a single, unique solution.