Inspiration
Privacy matters. Most AI coding tools send your code to remote servers — guIDE was built to change that. The goal: a fully capable, agentic IDE that runs entirely on your local hardware.
What it does
guIDE is a native desktop IDE built from the ground up for local LLM inference. Run AI models directly on your hardware — completely free for local models, no subscription or account required.
Key features:
- 69 built-in agentic tools — full tool loop for file operations, code execution, web browsing
- Playwright browser automation — the AI can actually browse the web
- RAG codebase indexing — context-aware responses about your entire project
- Code runner for 50+ languages — executes code directly
- Whisper voice input — dictate to the AI
- Full Git integration — commits, diffs, branches
- Privacy-first — your data never leaves your machine
How we built it
Built with Electron, C++, Node.js, Playwright, and Whisper. Uses a local LLM inference backend that runs entirely on-device.
Challenges we ran into
Making a truly offline AI tool that's still powerful enough to handle real agentic workflows — file operations, code execution, browser automation — all while keeping performance acceptable on consumer hardware.
What we learned
Local LLMs are ready for serious developer tooling. With the right architecture, you can build a capable agentic coding assistant that runs 100% offline.
What's next
Expanding RAG capabilities, adding more language support, and building out the model management UI.
Log in or sign up for Devpost to join the conversation.