Validate your hackathon idea against 320 real Google Gemini projects.
Stop building what already exists. Explore what's been done, find the gaps, and get ready-to-deploy prompts for v0, Lovable, and Google AI Studio—all in 90 minutes.
🚀 Try It Now • Report Bug • Request Feature
I participated in a hackathon with 320+ Google Gemini projects submitted. Analyzing those submissions manually was impossible.
So I built protoi — a tool that:
- Lets you search and explore all 320 projects instantly
- Uses AI to identify gaps and unbuilt ideas
- Generates deployment-ready prompts for modern builder tools
If you're about to start a hackathon project, this dataset is your competitive advantage.
Latest iteration featuring Flash UI integration with streaming code preview, advanced builder pipeline, and enhanced user experience.
📸 View Previous Versions
Previous versions maintained for reference
Search by technology, track, or AI capabilities. See what people actually built with Gemini.
Filters:
- 🧠 Complex Problem Solving (reasoning-heavy apps)
- 👁️ Image & Video AI (multimodal vision)
- ⚡ Real-Time Apps (live streaming, low latency)
- 🗣️ Voice & Audio AI (speech, TTS/STT)
- 📚 Long-Context Analysis (RAG, document processing)
- 🛠️ AI Agents & Tools (function calling, grounding)
Natural language chat with the entire dataset. Examples:
- "What healthcare problems haven't been solved?"
- "Show me projects that failed at monetization"
- "What's missing in the education space?"
Guided workflow from idea to deployment-ready prompt:
- Choose Platform → v0.dev, Lovable, or Google AI Studio
- Describe Idea → Natural language or "surprise me" mode
- Validate → AI checks if it's been done, finds similar projects
- Generate PRD → High-level product requirements document
- Get Prompt → Copy-paste ready prompt for your chosen platform
Time estimates:
- v0.dev: Working prototype in < 1 hour
- Lovable: Full app with auth in 1-2 hours
- Google AI Studio: Advanced AI demo in 1-3 hours
# Clone the repo
git clone https://github.com/Atypical-Playworks/protoi.git
cd protoi
# Install dependencies
pnpm install
# (Optional) Add your Gemini API key for AI Lab features
cp .env.example .env
# Start dev server
pnpm devVisit http://localhost:5173
Note: You can explore all 320 projects without an API key. The Gemini API is only needed for the "AI Lab" chat and builder features.
320 projects from the Google Gemini API Developer Competition, including:
- Project name, description, and track
- Technologies used (TensorFlow, React, Python, etc.)
- Team size
- Gemini capabilities leveraged (vision, audio, reasoning, context window)
- Links to demos, GitHub repos, and YouTube videos (when available)
Most popular tracks:
- Health & Medical AI
- Education & Learning Tools
- Creative & Content Generation
- Developer Tools & Productivity
Most used technologies:
- Gemini 1.5 Pro/Flash
- React/Next.js
- Python
- TypeScript
- TensorFlow
- Dataset loads automatically (320 projects cached locally)
- Use search bar or capability filters to find relevant projects
- Click any project to see full details
- Go to Build Mode (AI Lab)
- Choose "Ask The Data" to discover gaps
- Switch to "Build Your Project" when ready
- Follow the 5-step pipeline: Platform → Idea → Validation → PRD → Prompt
- Copy your deployment prompt and paste it into v0, Lovable, or AI Studio
Built for speed and simplicity:
- React 19 + TypeScript
- Vite 6 (dev server)
- Tailwind CSS (neumorphic design system)
- Google Gemini API (AI analysis)
- Recharts (data visualization)
- PapaParse (CSV parsing)
Found a bug? Have an idea? Contributions welcome!
- Fork the repo
- Create a feature branch (
git checkout -b feature/amazing-idea) - Commit your changes (
git commit -m 'Add amazing feature') - Push to the branch (
git push origin feature/amazing-idea) - Open a Pull Request
MIT License - Free and open source.
See LICENSE for details.
Built for those who need it. By someone who did.
I competed in the Gemini 3 AI Studio Hackathon. After submission, I discovered that 320 projects had been submitted—but they weren't publicly accessible until the hackathon ended. Once they were released, I wanted to analyze my competition and learn from every submission. The problem? Manually reviewing 320 projects was impossible. So I built a scraper to collect all the data, then built protoi to make sense of it all. I don't know if I'll be among the winners, but with protoi, I'm walking away with insights from all 320 competitors.
A project by Atypical-Playworks
Developed by Milumon