{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/ec0b6de82f7b4cb09d7c998bf3a7ef9f\" frameborder=\"0\" width=\"1470\" height=\"1102\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":1102,"width":1470,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":1102,"thumbnail_width":1470,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/ec0b6de82f7b4cb09d7c998bf3a7ef9f-6a70fd38964cbe69.gif","duration":641.64,"title":"Building AI Agents, From Recipes to Teams","description":"This Loom explains what AI agents are and how they differ from simple automation using a “chef” mental model. It describes an agent’s anatomy as context and state, tools and APIs, planning and routing, and an underlying LLM as the reasoning engine. The speaker then outlines three agent types: deterministic automation (fixed steps with limited LLM use), react style agents (reason and act within guardrails), and multi agent systems (multiple specialists coordinated together). The core principles emphasized are modularity, autonomy as a spectrum, and building with guardrails plus feedback and iteration."}