<?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/2d2b0b324e0941999508649a246c5481&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/2d2b0b324e0941999508649a246c5481-783590c44f1bef2e.gif</thumbnail_url><duration>300.099</duration><title>Present Flow: AI Slide Generation Architecture and Workflow</title><description>This Loom presents Present Flow, an automation project for generating editable PowerPoint decks for content weight assessment from written preferences. The core decision is to have the AI output structured slide data and a DEX specification rather than generating a PowerPoint file directly, keeping AI generation separate from slide rendering. The workflow uses Claude to send briefs for validation, then Anitan validates requests and uses OpenAI to produce structured deck specs, with a Step Renderer that validates and uses pptxgenjs to generate downloadable PPTX links. It demonstrates both a web front end and an MCP server pipeline that share the same workflow and annotation, including a longer render timeout to avoid cold start issues, and shows successful deck generation with returned presentation IDs and URLs.</description></oembed>