Design for the future of apps
summary
A demo called Gen Latte uses Gemini and Nano Banana to generate images then prints them on real lattes proving Gen UI can be end to end the team builds both a level two app and a level three party planner called Hatcha that lets an agent play 20 questions to compose an event UI the presentation names a four tier spectrum for Gen UI and an A2UI protocol for agent driven interfaces details include measuring every pixel to avoid ellipses token costs and caching generated screens to save money
AI-written summary · about
chapters (10)
- 0:00Introduction
- 1:03Flutter and open source
- 2:27Introducing Gen Latte
- 4:12What is GenUI
- 6:41GenUI adoption tiers
- 9:29AI-driven server UI
- 12:15AI take the wheel UIs
- 15:30Design challenges level 3
- 18:05Future of apps with GenUI
- 19:09Closing and Q&A
related talks
38mGenerative UIs: Using LLMs to Create Flutter Screens at Runtime
43mNext Steps and New Patterns for GenUI
28mLiquid User Interfaces for the Next Generation of AI Apps
39mAI-Powered Product Design for Bitcoin Experiences
01h 01mLessons Learned Building a GenAI Powered App
01h 03mBuild & deploy AI-powered apps
29mTools and Roadmap for AI-Native Developers
12mYour UI Has Left the Building: Interfaces That Write Themselves