<?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/8cd5a15710cb4f1894dbfb10dc6c1ae9&quot; frameborder=&quot;0&quot; width=&quot;1840&quot; height=&quot;1380&quot; webkitallowfullscreen mozallowfullscreen allowfullscreen&gt;&lt;/iframe&gt;</html><height>1380</height><width>1840</width><provider_name>Loom</provider_name><provider_url>https://www.loom.com</provider_url><thumbnail_height>1380</thumbnail_height><thumbnail_width>1840</thumbnail_width><thumbnail_url>https://cdn.loom.com/sessions/thumbnails/8cd5a15710cb4f1894dbfb10dc6c1ae9-aa60f260b993f048.gif</thumbnail_url><duration>25.804</duration><title>Noon - The Noon Report</title><description>Noon reconciles ad spend (Google, Meta, Taboola, TikTok) against real affiliate revenue into one honest profit-per-offer view, and shows a media buyer exactly where to move money each morning.

This demo runs the real reconciliation pipeline on synthetic data shaped like real platform exports — nothing hardcoded. The gap headline (platforms claim $154K, actually made $68K), the ranked money moves, and the live Refresh are all produced by the actual code. 91 tests green, including a golden eval that proves the pipeline&apos;s decisions against known truth.

Repo: github.com/ShaelaToni/operation-midday
Live URL (yeenom.com) deploying within the judging window; demo runs on synthetic data.</description></oembed>