Building AI-enhanced User Interfaces for the Web
summary
Builds a working, real-time collaborative AI-assisted editor in TypeScript that actually syncs AI and human edits using Yjs CRDT documents, a BFF, and a vector DB (pinecone) for context. Shows JSON/YAML streaming to render partial outputs, token-by-token streaming to keep latency under a few seconds, and tiny local models to fake responsiveness. Bots can write into the shared document, AI-generated words are tracked separately, edits remain non‑destructive and undoable, and the system demonstrably works.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 5:51AI writing app demo
- 6:55Large language models explained
- 17:45Human-AI collaboration modes
- 21:58UX for AI interfaces
- 29:05Natural language challenges
- 33:21AI collaboration architecture
- 40:00AI ethics and future
- 42:48Q&A
related talks
37mBuilding AI-powered user interfaces
35mBuilding AI-powered user interfaces
56mBeyond Chat: Bringing Models to the Canvas
28mCreating fluent Frontend Experiences in an age of AI
23mHuman-Centered AI: Supporting, Not Replacing, Frontend Developers
44mUnleashing the Power of AI: Integrating Tiptap, OpenAI, and Vue.js
19mBeyond agents: AI as a creative partner
45mWeb Development in the Age of AI