Prompt to Pipeline: Building with Google's Gen Media Stack
summary
Feeds Wind in the Willows through Gemini to produce character portraits with Nano Banana, animate scenes with VO and score chapters with LIA, then stitches them in Genie 3 into a 60‑second playable world with a pink sparkly squirrel that actually moves. AI Studio’s Build generates a bookshelf app with Google sign‑in and Firestore from one prompt; Gemini 3.1 Flash Light analyzes video frame‑by‑frame for under a dollar; Gemma 4 runs 10 sub‑agents on a local 26B to generate SVGs and open code debugs a game with no cloud API.
AI-written summary · about
chapters (19)
- 0:00Open source to Google DeepMind
- 7:57Morning recap and AI Studio demo
- 12:07Video analysis with Gemini 3.1 Flashlight
- 21:26Bookshelf app with AI Studio Build
- 24:13Live interaction with Gemini Live
- 29:09Exploring Genie 3 world model
- 35:12Deploying bookshelf app with Firebase
- 42:42Generative media models intro
- 51:35Book illustration with Nano Banana and VO
- 1:02:57Music generation with LIA 3
- 1:11:03Multi-voice audio with text-to-speech
- 1:14:01App from Python notebook
- 1:24:41Gemma 4 models and capabilities
- 1:31:48Running Gemma 4 locally
- 1:35:02SVGs with local Gemma agents
- 1:39:16Game with open code and Gemma 31B
- 1:49:06Webpage from screenshot
- 1:51:13Robotics and Android demos
- 1:54:04Q&A
related talks
27mBuild creative apps with the GenMedia suite
37mBuild agents with Gemini API
01h 45mAI Engineering with the Google Gemini 2.5 Model Family
49mAbracadabra! Bring Your Ideas to Life with Google AI Studio
18mIntroduction to Gemini APIs and AI Studio
29mTools and Roadmap for AI-Native Developers
15mBuilding with the Gemini API and AI Studio
21mAdd Generative AI capabilities to your Android app