Surviving the AI Bubble: The Open Source Guide from the Future
summary
The talk explains how to build a self-hosted open-source ChatGPT-style system on a consumer laptop, covering which open models and hardware to use and how to leverage smaller models. It demonstrates open Web UIs (Open WebUI, LM Studio, AnythingLLM) with RAG, shows agentic frameworks (LangChain, Mastra) to create custom agents and code assistants, and describes using LLMs for decision/classification to automate workflows (n8n, Make) while replacing generative fine-tuning and prompts with agentic architectures and embedding-based clustering to reduce dependence on commercial generative AI.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 1:57AI bubble burst in 2026
- 3:49Surviving without ChatGPT
- 4:09Self-hosting AI models
- 6:46Local AI with Ollama demo
- 8:09ChatGPT-like interfaces
- 9:57Open source AI agent platforms
- 11:46Open source code agents
- 12:50Office automation agent demo
- 14:39AI that works, not just talks
- 16:01Structured data extraction demo
- 17:33Custom AI agents
- 18:16Future of open source AI
