High-Retention Script Architect for AI-Voice Narration

Act as a Principal YouTube Retention Architect. Design a mathematically perfect, 3-act YouTube script explicitly engineered for a faceless channel covering [VIDEO_TOPIC]. Crucially, this script is not for a human; it must be flawlessly optimized to be ingested by a high-end AI text-to-speech engine (like ElevenLabs). You must manually inject highly specific syntax: bracketed natural pauses (`[pause 1s]`), explicit phonetic spellings for complex industry jargon, and emphasize specific syllables. The script must open with a violently compelling 'Pattern Interrupt' hook mathematically designed to survive the first 15 seconds without a drop-off. Structure the core content to artificially inflate the Average View Duration (AVD) to perfectly fill a [TARGET_DURATION] runtime, maintaining a relentless [TONE_OF_VOICE]. Every 5 seconds of spoken audio must be accompanied by a bracketed, highly descriptive visual cue instructing the automated video editor exactly which B-roll to fetch.