<?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/7ec7142420464242a26adc527927e585&quot; frameborder=&quot;0&quot; width=&quot;1108&quot; height=&quot;831&quot; webkitallowfullscreen mozallowfullscreen allowfullscreen&gt;&lt;/iframe&gt;</html><height>831</height><width>1108</width><provider_name>Loom</provider_name><provider_url>https://www.loom.com</provider_url><thumbnail_height>831</thumbnail_height><thumbnail_width>1108</thumbnail_width><thumbnail_url>https://cdn.loom.com/sessions/thumbnails/7ec7142420464242a26adc527927e585-1daca2deea9c9d9c.gif</thumbnail_url><duration>325.116</duration><title>Post implementation pt3: agentic QA workflow</title><description>This Loom explains what happens after delivery and how the author is using a swarm of QA agents to handle major bugs in a deck generation feature. They are delivering on staging with a production environment connected to it, and right now they are improving the study analysis section that produces cross-study overviews and external slide decks. Because fixing numerous visual and editing failures would take days alone, they informed a build orchestrator to run 10 QA agents plus an initial mapping agent using Playwright to review the deck in a browser and produce a detailed report. The report includes 3 validation failures and 15 successes, along with cost calculations to support modeling costs for each AI feature.</description></oembed>