I Stopped Using OpenClaw (and Built My Own) скачать в хорошем качестве

I Stopped Using OpenClaw (and Built My Own)

I spent 10 hours trying to get OpenClaw working on my Mac Mini M4 (bought last year), only to wait 2.5 minutes for a single response. So, I built my own version from scratch in less time than it took to setup OpenClaw. Introducing Gerald: A fully local, self-hosted AI agent built on n8n, running Qwen and Llama models entirely on my home network. No subscriptions, no data leaving my network, and no £800/month API bills. ✅ WHAT GERALD CAN DO: Control Linux servers via SSH (Full sudo commands) Live Wikipedia research for up-to-date facts Smart Home control via Home Assistant (Heating, lights, etc.) Local Permanent Memory using a vector database Near-instant responses with zero data leakage 🚀 CHAPTERS: 00:00 - The Mac Mini M4 & OpenClaw Hype 00:17 - Why I Quit OpenClaw (Security & Speed) 01:03 - The £800/Month API Problem 02:08 - Introducing Gerald: My Local AI Agent 02:49 - Demo: Linux SSH & Live Research 04:39 - Smart Home & Heating Control 06:14 - The Tech Stack: n8n, Llama & Qwen 11:36 - Home Assistant & Remote Access 12:23 - Future Plans: Images & MCP 14:05 - Final Thoughts: Build vs. Buy If you're wondering if OpenClaw is worth the struggle or the API costs, watch this first. #SelfHosting #HomeLab #LocalAI #OpenClaw #n8n #MacMiniM4 #AIAgent #Privacy #LLM #SmartHome