<?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/f0c96f8f162b46ed92417502580acf85&quot; frameborder=&quot;0&quot; width=&quot;1368&quot; height=&quot;1026&quot; webkitallowfullscreen mozallowfullscreen allowfullscreen&gt;&lt;/iframe&gt;</html><height>1026</height><width>1368</width><provider_name>Loom</provider_name><provider_url>https://www.loom.com</provider_url><thumbnail_height>1026</thumbnail_height><thumbnail_width>1368</thumbnail_width><thumbnail_url>https://cdn.loom.com/sessions/thumbnails/f0c96f8f162b46ed92417502580acf85-3e8315efad87dec6.gif</thumbnail_url><duration>192.904</duration><title>Ops Assist AI for Manufacturing Troubleshooting</title><description>This Loom introduces Ops Assist, an AI-powered manufacturing troubleshooting assistant built using Jemma4 for the Dev Jemma4 Challenge. It lets operators enter machine symptoms, alarms, or issue descriptions in natural language, then uses the Jemma4 API to generate likely causes, recommended checks, safety reminders, and escalation recommendations. The example focuses on a conveyor belt that drifts sideways after stop and start cycles, causing jams on one lane, with no alarms and a low severity observation. It also explains that the app uses the gemma 4 26BA4B model to balance strong reasoning performance with fast response time.</description></oembed>