Claire Vo built a complete AI avatar video of herself in 15 minutes using Google Flow and the Gemini Omni video generation model. The workflow: phone scan to capture her face, AI-generated storyboard, seven distinct scenes with character consistency locked across all of them, then stitched together in a browser-based editor. No production crew, no video skills required.
The real value here is not the finished video. It is watching the process fail in real time. Vo accidentally generates static images instead of video clips at the 8:41 mark, hits uncanny valley on emotional expressions and basic physics, and works through it anyway. That troubleshooting sequence is worth more than any polished tutorial because it shows exactly where Gemini Omni breaks down in 2025.
Google Flow, Veo 3, and Gemini Omni are the three tools doing the heavy lifting here. All are live and linkable now. The question the full episode forces you to sit with: if a one-person team can produce a credible hype video in under 15 minutes today, what does that make a video production budget worth in 12 months.
[READ ORIGINAL →]