Steve Yegge, veteran of Amazon and Google and author of the new book 'Vibe Coding', argues that hand-written code will gradually disappear. In a conversation recorded at Martin Fowler's Future of Software Development workshop in Utah in early February, Yegge lays out a framework of AI adoption levels for engineers, from avoiding AI tools entirely to running multiple agents in parallel. He also built an open-source AI agent orchestrator called Gas Town and explicitly tells most developers not to use it.
The episode almost did not exist. Poor audio from the original recording forced host Gergely Orosz to publish a written deep-dive first. Engineer Tatsuhiko Miyagawa and post-production tool Auphonic salvaged the raw audio, and the full conversation is now available, including sections the written version cut. That material, covering Yegge's unfiltered takes on technical debt acceleration, productivity pressure, and what knowledge actually stays relevant as tooling shifts, is the reason to watch rather than just read the summary.
Yegge's core claim is that AI amplifies engineers rather than replacing them, but he pairs that with a direct warning: the pace of change will introduce new categories of technical debt and reshape how teams operate. The transcript and timestamped chapters are available alongside the video on YouTube, Spotify, and Apple Podcasts. Read the original for the specific level-by-level breakdown of AI adoption and Yegge's reasoning on why systems-level thinking outlasts any particular tool.
[READ ORIGINAL →]