Phil and Rohan go deep on practical AI for Home Assistant. We cover LLM Vision for doorbell and perimeter alerts, HACS add‑ons like AI Agent and AI Suggester, and using GitHub Copilot to generate YAML automations. Rohan shares his Ollama + LiteLLM routing setup and why private cloud models might beat pay‑per‑token APIs. Phil battles OpenClaw (aka MoltBot/Claude Bot) to control lights via MCP
Links Discussed
- https://github.com/sbenodiz/ai_agent_ha
- https://github.com/valentinfrlch/ha-llmvision
- https://github.com/ITSpecialist111/ai_automation_suggester
Chapters
00:00:00 – Intro
00:02:02 – Rohan’s AI stack today
00:09:00 – HA’s built-in AI + Copilot
00:18:00 – HACS: AI Agent & Suggester
00:26:24 – LLM Vision in practice
00:37:00 – OpenClaw, MCP, and gotchas
00:47:00 – Do we need agents? Use cases
00:51:00 – Costs, feedback, and outro
00:02:02 – Rohan’s AI stack today
00:09:00 – HA’s built-in AI + Copilot
00:18:00 – HACS: AI Agent & Suggester
00:26:24 – LLM Vision in practice
00:37:00 – OpenClaw, MCP, and gotchas
00:47:00 – Do we need agents? Use cases
00:51:00 – Costs, feedback, and outro
