Your Personal AI Agent. On Your Device. On Your Terms.
Giga Bot is a LOCAL-FIRST desktop application that wraps and controls OpenClaw, making powerful AI agents accessible without technical setup.
Giga Bot is NOT an AI runtime. It's a wrapper and controller that:
- ✅ Installs OpenClaw from GitHub automatically
- ✅ Detects your device capabilities intelligently
- ✅ Manages OpenClaw lifecycle (start/stop/restart)
- ✅ Provides a beautiful UI for chat and tasks
- ✅ Routes between local and cloud AI seamlessly
- ✅ Supports Telegram, WhatsApp, and Web
OpenClaw runs locally on your device. Giga Bot just makes it easy.
┌─────────────────────────────────────────────────┐
│ Giga Bot Desktop (Electron) │
│ │
│ ┌─────────────────────────────────────────┐ │
│ │ Device Intelligence Engine │ │
│ │ - Hardware detection │ │
│ │ - Capability scoring │ │
│ │ - LLM routing decisions │ │
│ └─────────────────────────────────────────┘ │
│ │
│ ┌─────────────────────────────────────────┐ │
│ │ OpenClaw Installer │ │
│ │ - GitHub clone │ │
│ │ - Dependency installation │ │
│ │ - Validation │ │
│ └─────────────────────────────────────────┘ │
│ │
│ ┌─────────────────────────────────────────┐ │
│ │ OpenClaw Manager │ │
│ │ - Process lifecycle │ │
│ │ - Health monitoring │ │
│ │ - Auto-restart on crash │ │
│ └─────────────────────────────────────────┘ │
│ │
│ ┌─────────────────────────────────────────┐ │
│ │ Gateway WebSocket Client │ │
│ │ - Bidirectional communication │ │
│ │ - Message streaming │ │
│ │ - Tool invocation │ │
│ └─────────────────────────────────────────┘ │
└─────────────────────────────────────────────────┘
│
│ WebSocket (localhost:18789)
▼
┌─────────────────────────────────────────────────┐
│ OpenClaw Runtime (Local) │
│ Installed from GitHub │
│ ~/.openclaw/ │
└─────────────────────────────────────────────────┘
gigabot-desktop/
├── src/
│ ├── main/
│ │ └── main.ts # Electron main process
│ ├── preload/
│ │ └── preload.ts # IPC bridge
│ ├── renderer/
│ │ └── (UI components) # React UI
│ ├── modules/
│ │ ├── device-intelligence/
│ │ │ └── DeviceIntelligenceEngine.ts
│ │ ├── openclaw-installer/
│ │ │ └── OpenClawInstaller.ts
│ │ ├── openclaw-manager/
│ │ │ └── OpenClawManager.ts
│ │ └── gateway-client/
│ │ └── GatewayClient.ts
│ └── shared/
│ └── (shared types)
├── package.json
├── tsconfig.json
└── README.md
- Node.js 18+ (will be auto-installed if missing)
- pnpm (recommended) or npm
pnpm installpnpm devpnpm buildThis creates platform-specific installers in release/:
- macOS:
.dmgand.zip - Windows:
.exeinstaller and portable - Linux:
.AppImageand.deb
Automatically detects:
- OS, CPU, RAM, GPU, NPU
- Disk space availability
- Device Ready Score (Low/Medium/High)
- Recommended local models
User sees:
"Your device is perfect for running AI locally!"
Not:
"16GB RAM, NVIDIA RTX 3060 with 8GB VRAM detected"
On first launch:
- Detects device capabilities (5-10 seconds)
- Installs OpenClaw from GitHub
- Installs Node.js (if missing)
- Installs Ollama (if device supports local AI)
- Downloads recommended local model
- Validates everything
User sees:
"Preparing your device for personal AI (2-4 minutes)"
Decides local vs. cloud based on:
- Device capability score
- Task complexity
- Context size
- User preference (Private Mode / Cloud Boost)
Never exposes:
- Token counts
- VRAM usage
- Model quantization details
- Monitors OpenClaw process health
- Auto-restarts on crash (configurable max attempts)
- Captures logs for debugging
- Graceful shutdown handling
OpenClaw configuration is generated automatically based on:
- Device capabilities
- User preferences
- Available API keys
Configuration is stored in:
~/.openclaw/config.json
// Detect device capabilities
const profile = await window.gigabot.device.detect();
console.log(profile.capabilities.deviceReadyScore); // 'high' | 'medium' | 'low'
console.log(profile.capabilities.userMessage); // User-friendly message
console.log(profile.capabilities.recommendations); // Array of recommendations// Install OpenClaw
await window.gigabot.openclaw.install();
// Start OpenClaw
await window.gigabot.openclaw.start();
// Check health
const health = await window.gigabot.openclaw.getHealth();
console.log(health.status); // 'running' | 'stopped' | 'error'
// Get logs
const logs = await window.gigabot.openclaw.getLogs(100);// Send message
await window.gigabot.chat.send({
message: 'Hello, world!',
sessionId: 'session-123',
});
// Listen for responses
window.gigabot.chat.onMessage((message) => {
console.log(message.content);
});- Local-first: OpenClaw runs on your device
- No silent cloud execution: User must explicitly enable Cloud Boost
- Encrypted credentials: API keys encrypted at rest (AES-256)
- Open source: Full transparency, no hidden telemetry
MIT License - see LICENSE
Built on top of OpenClaw - the open-source AI agent framework.
Giga Bot is a wrapper, not a fork. We contribute upstream and benefit from community improvements.
- Phase 1: Device Intelligence Engine
- Phase 2: OpenClaw Installer & Manager
- Phase 3: Gateway WebSocket Client
- Phase 4: Local LLM Orchestration (Ollama)
- Phase 5: Desktop UI (React)
- Phase 6: Messaging Platform Integrations
- Phase 7: Optional Cloud Sync
Built with ❤️ by Gignaati