<?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/97c4f3df8f5b4642aabd1195b94a67a1&quot; frameborder=&quot;0&quot; width=&quot;1920&quot; height=&quot;1440&quot; webkitallowfullscreen mozallowfullscreen allowfullscreen&gt;&lt;/iframe&gt;</html><height>1440</height><width>1920</width><provider_name>Loom</provider_name><provider_url>https://www.loom.com</provider_url><thumbnail_height>1440</thumbnail_height><thumbnail_width>1920</thumbnail_width><thumbnail_url>https://cdn.loom.com/sessions/thumbnails/97c4f3df8f5b4642aabd1195b94a67a1-b9cc017b5166919b.gif</thumbnail_url><duration>251.427</duration><title>Preventing AI Hallucination in Code Outreach with Human-in-Loop Workflow ✉️</title><description>Hi, I’m Abdul and in this video, I demonstrate an AISDR system I built that incorporates a human-in-loop workflow to prevent AI hallucination in code outreach. I walk you through two scenarios where we fetch a lead, draft a personalized email using the OpenRouter AI API, and send it for approval via a custom Slack app. After running the workflow, I show how the email was successfully sent to my Gmail after approval. I also explain how I utilized a rejects text parser to clean up the AI-generated content and added a brick arrow handler for system reliability. Please let me know your thoughts on this workflow and if you have any suggestions for improvement.</description></oembed>