Next Steps and New Patterns for GenUI
summary
Explains how GenUI actually delivers UI: generating and saving A2UI messages from Gemini into Firestore via Dart functions, then loading those precached messages into the app to avoid wait times; shows combining agentic state with local component state to reduce hallucination by sending only job IDs and fetching authoritative data clientside; demos client side functions for deterministic arithmetic and a server trigger that precomputes feeds so generated UI appears instantly and works.
AI-written summary · about
chapters (9)
- 0:00GenUI and ATUI protocol
- 3:28From demos to GenUI apps
- 7:17Structured output and GenUI integration
- 10:09Pre-generating ATUI messages
- 14:25Dart functions for Firebase triggers
- 25:09Reducing hallucinations with client data
- 26:31Client-side functions in GenUI
- 33:45GenUI and ATUI resources
- 35:55Q&A on caching
related talks
38mGenerative UIs: Using LLMs to Create Flutter Screens at Runtime
20mDesign for the future of apps
01h 01mLessons Learned Building a GenAI Powered App
27mSmarter UIs: Building Context-Aware AI Apps with Genkit & Angular
32mGenerative UI in Flutter
46mLessons Learned Building A GenAI Powered App
36mEmpowering Your Flutter Apps with AI
39mA2UI for Android: Safely Rendering AI-Generated UI with Jetpac…