{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/f3479c6b94034ca29f376803726d2299\" frameborder=\"0\" width=\"1248\" height=\"936\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":936,"width":1248,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":936,"thumbnail_width":1248,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/f3479c6b94034ca29f376803726d2299-38ae4b4ef63cec10.gif","duration":371.916667,"title":"NEW: AI Agent Skills Walkthrough 🎥","description":"This Loom explains why the product team is introducing skills to improve how AI agents follow standard operating procedures. Felipe describes how guidance-based processes can be inconsistent, such as agents prioritizing sent articles or website information over the intended steps for tasks like returns, order tracking, and order edits before checkout. Skills, built with instructions plus intent-based triggers, aim to deliver consistent behavior whenever an order cancellation intent is detected. The Loom also covers knowledge toggles and converting procedural guidance into a skill, and introduces a view of intent traffic where specific intents can be turned off for human handling."}