<?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/fa0d160cc9e147a4ad4c8128a3c909d2&quot; frameborder=&quot;0&quot; width=&quot;1110&quot; height=&quot;832&quot; webkitallowfullscreen mozallowfullscreen allowfullscreen&gt;&lt;/iframe&gt;</html><height>832</height><width>1110</width><provider_name>Loom</provider_name><provider_url>https://www.loom.com</provider_url><thumbnail_height>832</thumbnail_height><thumbnail_width>1110</thumbnail_width><thumbnail_url>https://cdn.loom.com/sessions/thumbnails/fa0d160cc9e147a4ad4c8128a3c909d2-c11b6a7ccaaed8ed.gif</thumbnail_url><duration>64.463</duration><title>Project Call Autopsy for AI Voice Agents</title><description>This Loom explains an “autopsy” pipeline for debugging voice agent conversations when an agent drops or fails to answer. It describes how the system analyzes an entire VAPI or bot conversation after a failure to identify exactly where the bot went wrong. The goal is to produce a detailed report pinpointing the issue and automatically apply fixes to resolve the problem.</description></oembed>