The Bitter Layout or: How I Learned to Love the Model Picker
summary
Argues that the ubiquitous input field + turn‑by‑turn flow + dropdown of models—the model picker—exists because models are not yet commoditized, citing Christensen’s modular vs integrated architecture and Sutton’s bitter lesson to claim the UI must conform to the next model. Uses examples across chatbots, coding assistants (Vzero) and Canva, shows a demo of an older ChatGPT model mismatch, and proposes shifting to goal/constraint design, design systems as prompts, and RL‑style QA.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 0:23The Idea
- 1:42The Great Chatbot Debate
- 3:29Models and Modes
- 5:45Flexibility Usability
- 6:30The zeitgeist
- 6:54Product architecture
- 7:26Types of architectures
- 8:52The Bitter Lesson
- 9:27Bitter Design Lesson
- 10:10Why the Bitter Layout Works
- 10:46The Takeaway
- 11:19The Future of Programming
- 12:11Design Systems
related talks
56mBeyond Chat: Bringing Models to the Canvas
46mDesign Patterns For AI Interfaces
27mThe Shape of AI : How designers integrate AI into their workflow
20mBridging the gap between AI and UI
50mKEYNOTE: The agentification of software has a UX problem
17mBeyond Components: Designing Generative UI for MCP Apps
45mModeling as Mind Redesign: Why participatory AI is better
29mBuilding Scalable and Adaptive Products in a Fast Evolving Landscape