Descriptions:
In this episode of How I AI, host Claire walks through a live, end-to-end experiment using Google Flow and the Gemini Omni video generation model to create a personalized AI avatar of herself — from scanning a QR code for facial capture to generating finished video clips in under 15 minutes.
The video documents the full workflow: capturing reference photos via Google Flow’s avatar tool, prompting Gemini Omni to storyboard a hype video for her podcast, iterating on scene descriptions (extreme close-ups at a mechanical keyboard, a dark-green hacker-aesthetic home office), and troubleshooting a common mistake of accidentally selecting image generation instead of video. Scenes are queued in parallel and intended for final stitching.
Claire reflects on what multimodal AI tooling unlocks for solo creators: the ability to produce polished, character-consistent video content — storyboarding, shot blocking, avatar animation — without a production team. For anyone exploring Google Flow, Gemini Omni’s avatar creation feature, or the practical limits of AI-generated video for personal branding and podcast promotion, this video is an honest first-person account of what works, what fails on the first attempt, and what the results actually look like.
📺 Source: How I AI · Published June 03, 2026
🏷️ Format: Hands On Build
