The Four Horsemen of Agentic Coding
Agentic coding creates four major systemic harms for engineering teams: unmaintainable code slop, engineer alienation, widespread deskilling, and the destruction of team social fabric.
read it It articulates the unspoken psychological and cultural downsides of AI-driven development that many engineers are quietly experiencing.
summary
The article outlines four critical drawbacks of agentic coding: slop, alienation, deskilling, and team fallout. LLM-generated code creates repulsive codebases, widening the gap between engineers and their craft while eroding core technical skills. Furthermore, reliance on autonomous AI agents silences team chat channels and destroys the collaborative social fabric of engineering. Ultimately, while productivity may rise, the long-term impact threatens the craft and culture of software engineering.
key points
- LLM output often manifests as unmaintainable slop that repels humans from wanting to read or maintain codebases.
- Abstracting away the actual writing of code alienates engineers from their work, causing them to care less about quality and root-cause fixes.
- Relying entirely on AI easy buttons creates severe deskilling risks and breaks traditional skill progression paths for junior engineers.
- Engineering team communication collapses into isolated silos where individuals interact exclusively with their personal AI agents instead of peers.
who should read it
Software engineers, engineering managers, and technical leaders grappling with the cultural and quality impacts of AI coding tools.
discussion skeptical
consensus
Commenters overwhelmingly validate the author's observations, sharing anecdotes of ghost-town Slack channels, code review fatigue, and the erosion of collaborative architecture discussions.
pushback
A dissenting minority argues that agentic coding is immensely empowering and fast, and that these negative outcomes stem from poor individual workflows rather than inherent flaws in the tools.
My work Slack channels are ghost-towns. People are logged in, but they're never saying anything. I often find myself wondering if anyone is even working. PRs have their names on them, even the review comments are posted from _their_ accounts, but it is obvious that Claude wrote everything from the code, the unnecessarily verbose change summary, to their comments. Even their replies to my change requests start with "You're right — …".
I miss the days when the engineering Slack channels had long threads of new and interesting ways to do things, discussions of something we could do next to make …
It provides a vivid firsthand account of how engineering Slack channels have turned into ghost towns where even PR comments and responses are entirely generated by AI.
Generated by an LLM from the article and thread. Check the sources before relying on it.