Build AI Agents That Run Locally
GAIA is an open-source framework for building AI agents in Python and C++ that run on local hardware. Agents reason, call tools, search documents, and take action — the core runtime needs no cloud, so sensitive data stays on-device. Individual agents can still opt into external services (weather APIs, Jira, MCP servers) when a use case requires them.
Local Inference
All processing stays on-device
Cloud-Optional
Core runtime needs no cloud or API keys — agents can opt in when a use case requires it
Python & C++
Full SDK in both languages
AMD Optimized
NPU and GPU acceleration on Ryzen AI
- Python
- C++
Get Started
Agent UI
Install via npm and run AI agents locally — two commands to get started.
Python Quick Start
Install GAIA, start Lemonade Server, and run your first agent.
C++ Quickstart
Build a native C++17 agent binary — no Python runtime required.
Capabilities
- Python
- C++
Agent UI
Privacy-first desktop chat with drag-and-drop document Q&A. Install via npm or
gaia --ui.Document Q&A (RAG)
Retrieve, index, and answer questions over local PDFs, code, and text files.
Speech-to-Speech
Whisper ASR and Kokoro TTS — offline voice interaction pipeline.
Flagship GAIA Agent
Conversation, documents, data, web, memory, and skills in one agent.
Code Index
Semantic search over your codebase using local AMD-accelerated embeddings.
MCP Integration
Connect agents to external tools via Model Context Protocol.
Composing Skills
Extend the flagship agent’s capabilities with a
SKILL.md.Resources
SDK Reference
Complete API documentation
GitHub
View source code
Component Specs
Technical specifications
Glossary
Technical terms explained
Discord
Join community