<?xml version="1.0" encoding="UTF-8"?><oembed><type>video</type><version>1.0</version><html>&lt;iframe src=&quot;https://www.loom.com/embed/6e9842fb18a74efa8b12e4e0a8fcb8ab&quot; frameborder=&quot;0&quot; width=&quot;1646&quot; height=&quot;1234&quot; webkitallowfullscreen mozallowfullscreen allowfullscreen&gt;&lt;/iframe&gt;</html><height>1234</height><width>1646</width><provider_name>Loom</provider_name><provider_url>https://www.loom.com</provider_url><thumbnail_height>1234</thumbnail_height><thumbnail_width>1646</thumbnail_width><thumbnail_url>https://cdn.loom.com/sessions/thumbnails/6e9842fb18a74efa8b12e4e0a8fcb8ab-30ff277e695b5f83.gif</thumbnail_url><duration>136.237</duration><title>Innovative Multimodal Reasoning Agent for Welding Systems Explained 🔧</title><description>In this video, I present my multimodal reasoning agent designed for the Omni Pro 220 welding system. The core idea is to transform a 40-page manual into structured knowledge, including text notes, tables, procedures, diagrams, and their relationships. I demonstrate how the agent retrieves information and simulates machine states, providing visual artifacts instead of just text. For example, I show how it responds to queries about polarity in TIG welding and duty cycle specifications, rendering diagrams and thermal limits. I encourage you to explore the code base and request a live demo for a deeper understanding.</description></oembed>