Code is cheap. Now what?
summary
Claim: coding agents make code is cheaper, shifting decisions about legacy, architecture and buy vs build. Evidence: examples of cloning and reimplementation (Chardet clean‑room relicensed) and a runtime rewritten from Zigg to Rust in ~11 days (≈500k → 1M lines) using agents. Effects: proliferation of niche apps and rich prototypes that create prototype > product and validation debt, token budgets replacing some labor costs, uncertainty for junior hiring, and a push toward scale to zero ownership with serious operational risk.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 2:48Code becoming cheaper
- 7:14Niche software and prototypes
- 9:42Software development trends
- 12:32Labor versus tokens
- 13:54Future of junior roles
- 15:59Cloning software and moats
- 18:26Research as advantage
- 19:49Regenerating software
- 21:57Worth's Law and quality
- 24:47Zero ownership systems
- 28:17Jevons paradox in development
- 30:25Career advice in AI era
- 30:46Q&A
related talks
19mCode Is Cheap. Software Isn't
21mDesigning in the age of the 10x engineer
47mCode Is Cheap. Software Isn’t.
25mBeyond the IDE: Toward Multi-Agent Orchestration
35mUsing coding agents with open models
16mCoding Reinvented How AI Agents Are Shaping the Future of Software Development
16mBeyond the IDE: The Future of Developing Software With Agents
18mCoding Reinvented: How AI Agents Are Shaping the Future of Software Development