2025年11月18日LukeEnglish
Beyond Vibe Coding: Kent Beck on Augmented Coding
Kent Beck’s distinctions between Augmented Coding and Vibe Coding, plus his value system; English translation provided.
Sharing Kent Beck’s framing of Augmented Coding vs Vibe Coding, which aligns with what we pursue.
Original summary: https://tidyfirst.substack.com/p/augmented-coding-beyond-the-vibes
Kent Beck separates “Augmented Coding” and “Vibe Coding” based on how much the developer cares about code quality and control.
| Category | Vibe Coding | Augmented Coding |
|---|---|---|
| Care for code | Little concern for internal quality | Care about complexity, tests, and coverage |
| Error handling | Ask the “genie” for a good-enough fix | Watch interim results and step in when needed |
| Goal | Focus on behavior only | Production-quality, performance-competitive code |
| Developer role | Mostly delegate functionality | Lead design/TDD and use AI as support |
Augmented Coding vs Vibe Coding
Beck realized the split after narrowing scope to a B+ Tree library, then expanding again to see if AI could reach production-level performance.
Value system of Augmented Coding
- Tidy code that works remains the goal, even if AI types most of it.
- TDD & Tidy First: keep Red → Green → Refactor; separate structural (non-behavior) from behavioral changes—never mix them, and do structure first.
- High code quality: remove duplication, name/structure for clarity, choose the simplest solution.
- Commit discipline: commit only when warnings are gone, tests pass, and the change is a single logical unit.
Analogy
If “Vibe Coding” is asking an AI chef to wing a tasty meal without a recipe, “Augmented Coding” is the developer-chef strictly managing ingredients, cleanliness, and steps, asking the chef to follow TDD/Tidy First while offloading only the tedious parts. AI writes code, but quality and structure stay the developer’s responsibility.
More posts

In the AI-agent era, we believe what developers and enterprises need is not a comforting companion, but an equal partner that helps teams achieve outcomes. Here’s how we think about Caret and Caretive.

— Graph RAG, Knowledge Graphs, and Semantics-Based AI Architecture
