E1: Design-Centric Work
AI can generate code quickly but cannot judge "what should be built." If humans still spend most time writing code, they're doing what AI excels at while neglecting what AI cannot do well.
Let humans focus on work AI cannot replace: understanding requirements, making design decisions, verifying results are correct. The work distribution shifts from "Design 20%, Implementation 60%, Testing 20%" to "Design 40%, Implementation (AI-led) 20%, Review 40%."
Principle
- All roles MUST shift focus toward design and design review. Execution becomes increasingly delegated to AI agents.
Humans invest more time in defining what to build and verifying it's built correctly, rather than how to implement. Design areas include requirements design (defining problems, user needs, success criteria), architecture design (system components, data flow, technology selection), API design (interface contracts, data models, error handling), UX design (user flows, interface layouts, interaction patterns), and test design (test strategy, test cases, coverage scope).