The Rise of Malleable Software 🧩 — with Geoffrey Litt
Summary
Luca interviews Jeffrey Litt, a design engineer at Notion, about the concept of malleable software, which balances structure with user customization through progressive disclosure. They discuss how AI agents and Notion’s new developer platform allow users to build custom tools and code directly within the workspace. The conversation explores shifting role boundaries, enabling non-engineers to contribute via AI, and strategies for managing trust and oversight in automated workflows.
Topics discussed
Introduction and the concept of malleable software
Guest introduction and episode overview
Jeffrey Litt's background and the origin of malleable software
The tension between blank canvases and rigid apps
Sponsor segment: Unblocked AI context engine
The nightmare bicycle parable and primitive components
Progressive disclosure and the learning curve trade-off
Democratizing software creation with AI agents
Notion blocks as the fundamental primitive
Using Notion pages as rich context for AI agents
Vibe coding vs. platform-based development
Introducing Notion Workers for custom code
Notion's shift to a horizontal AI workspace
Non-developers shipping code and the 100% bar
Managing AI agents and tasks within Notion
Horizontal context and the OpenClaw comparison
Trust, oversight, and version control for agents
Reviewing agent work and human intervention
The design engineer role and rapid prototyping
Internal dogfooding, feature flags, and power users
Blurring the lines between design and engineering
Code as the primary source of truth for prototypes
Future roadmap and closing remarks
Listen ad-free on Castria