Google used its own AI stack to help produce much of Google I/O 2026. From the jellyfish pre-show to the Timmy TPU film, the company set itself the challenge of using the same Gemini tools it was putting on stage — to out-create and out-efficient itself.
The AI That Built the Show
The I/O brand identity was a close human-plus-AI collaboration. Googlers fed Gemini past brand guidelines and five years of I/O recaps, iterated with Nano Banana, and landed on a four-color gradient with interlocking icons that transform from flat 2D into hyper-textured 3D across the event.
AI x Film: Timmy TPU
The highlight was TPU Training Day, a short film made from cardboard, markers, puppetry, and AI. Teams captured character performances through puppetry and simple 3D animation, used Nano Banana to generate stylized first frames with pixel-perfect consistency via a custom Google AI Studio tool, and merged the base animation with generated sequences using Gemini Omni — deliberately preserving the tiny human imperfections that give puppet films their charm.
AI x Immersive Experiences
The pre-show pushed AI into interactive territory:
- Jellectronica — a YOLO8 model tracked jellyfish movement to control music generated with Lyria 3 Pro
- Infinite Scaler — players generated playable 3D worlds from 2D image prompts via Nano Banana and the Gemini API
- Code the Countdown — creators worldwide designed numbers in Canvas or AI Studio, stitched into a live countdown
AI x Creative Delight
Every speaker received a custom title card generated with Nano Banana Pro, with actions prototyped in Google Flow using Veo. Attendees could also catch falling prompts in a 20-second web-app game that fused their picks and printed custom I/O stickers on the spot — gold-waffle I/O logos and gummy-bear motherboards included.
The Point of Doing It
Google says the exercise was about showing what AI actually enables: offloading the mundane tasks so teams can spend their best hours on the parts only humans do well. When done right, viewers stop thinking about how AI was used — and that shift is the message.
What This Means
Using production tools to produce the flagship event is Google’s most public dogfooding yet. It is also a preview — if the world’s biggest developer conference can run this much of its pipeline on generative AI, the equivalent workloads inside every company just got smaller.