Inspiration
We wanted to make AI-powered personal operating systems accessible to everyone—even those with very limited hardware. Most AI OS solutions demand high compute and expensive APIs. Our goal: full AI-automation with 2GB RAM and $0 monthly cost.
What It Does
- Execute natural language commands: "Open WhatsApp and call Kartik", "Create a Python project", "Schedule a meeting tomorrow"
- Manage files, launch apps, and monitor processes
- Automate communications (WhatsApp, email, SMS, social media)
- Seamlessly integrate with VS Code, Git, and developer workflows
- Perform AI-powered reasoning and automation using Hugging Face free-tier APIs
How We Built It
We designed a modular MCP (Model Context Protocol) architecture, with a central orchestrator and multiple lightweight MCP servers for system ops, communication, browser automation, IDE integration, and AI execution.
Everything is optimized for 2GB RAM using:
- Lightweight Docker containers
- Lazy-loading MCP servers
- Free-tier API optimization (request batching, caching, fallback routing)
Challenges
- Running advanced AI reasoning on just 2GB RAM
- Managing Hugging Face free-tier limits (solved with smart routing + caching)
- Sandboxing Python execution securely in Docker
- Optimizing memory across multiple servers without crashes
Accomplishments
- Fully functional AI OS running on 2GB laptops and cloud VMs
- $0 monthly cost with Hugging Face, Groq, and Gemini free APIs
- Successfully executed 50+ natural language commands in testing
- Built web dashboard + CLI for easy user interaction
What We Learned
- Efficient API routing and memory management
- Building modular AI microservices with MCP
- Designing for extreme hardware constraints without compromising features
What’s Next
- Expanding marketplace for plug-and-play MCP servers
- More advanced multi-step planning and error recovery
- Production deployment with Kubernetes for teams
- Community-driven extensions and plugins
Built With
- docker
- docker-api-databases/tools:-git
- fastmcp
- gmail-api
- google-gemini-free-tier-apis:-hugging-face-gpt-oss
- groq-free-tier
- kubernetes
- languages:-python
- node.js
- notion-api
- react
- slack-api
- tauri-cloud-services:-hugging-face-inference-api
- typescript-frameworks:-fastapi
- vs-code-integration
- whatsapp-web-(selenium)
