<?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/816e8d7ab02f41a78b54f239f2724560&quot; frameborder=&quot;0&quot; width=&quot;1106&quot; height=&quot;829&quot; webkitallowfullscreen mozallowfullscreen allowfullscreen&gt;&lt;/iframe&gt;</html><height>829</height><width>1106</width><provider_name>Loom</provider_name><provider_url>https://www.loom.com</provider_url><thumbnail_height>829</thumbnail_height><thumbnail_width>1106</thumbnail_width><thumbnail_url>https://cdn.loom.com/sessions/thumbnails/816e8d7ab02f41a78b54f239f2724560-4ce9ef0fc5ec9e22.gif</thumbnail_url><duration>209.728</duration><title>GroundTruth — Evals for enterprise data</title><description>This Loom demonstrates Groundtruth, an autonomous steward agent that verifies, scores, and recommends changes to keep database data from becoming stale. The walkthrough shows uploading a CSV, previewing the data, and starting an evaluation that runs row by row while multiple agents work, including an ingestion agent for field mapping, a source agent for relevant fields, a control agent for monitoring and flagging, and a reporting component. In one completed project, evidence and agent trace indicated a link was incorrect and needed correction. The creator also mentions future plans for email tracking to request updated data from companies and CRM integration to bring CRM data into the system, built for Codex Lyathon 2.9.</description></oembed>