Generally Intelligent Malleable software and human agency with Geoffrey Litt
11 snips
Nov 14, 2025 Geoffrey Litt is a design engineer at Notion and a former MIT researcher focused on malleable software that lets users shape their own computing environments. He discusses barriers to malleability in mainstream software and the need for user-editable tools. Geoffrey explores AI's role in the creative process, emphasizing how it can enhance or hinder user agency. He also advocates for universal versioning and gradual tailorability, envisioning a future where users become active creators of their digital tools.
AI Snips
Chapters
Transcript
Episode notes
Smooth Slope From User To Creator
- Malleability isn't about everyone becoming full-stack experts but about maintaining a smooth slope from user to creator.
- Local power users and pattern local developers play key roles in enabling others to modify tools in context.
Use AI To Teach, Not Just Do
- Design AI features to teach and surface best practices proactively, not just auto-do tasks for users.
- Use suggestions and explainers to raise users' understanding while lowering the effort to act.
GUIs Need Composable Hooks
- GUIs excel at visualization and discoverability but often hide composable hooks that make automation easy.
- Command-line-style parity between manual and automated steps made composability obvious in older tools.