AI coding tools were supposed to make software engineers more productive. One engineer now argues they are doing the opposite, turning developers into little more than operators who spend their days approving machine-generated work.
Writing anonymously on X under the username voxium, the engineer described his job at an undisclosed company as “soul-sucking,” blaming the company’s heavy reliance on Anthropic’s Claude Code.
According to voxium, Claude is now being used to generate far more than source code. Product specifications, tests, tickets, reports, and other routine development work are increasingly being handed over to AI.
The result, he claims, is a workplace where engineers can spend 12 to 13 hours a day essentially reviewing and approving what the AI produces.
“Nobody Is Thinking Anymore”
I am done with this shit. It is over. The state of engineering right now is horrible. It has been half a month since I started a new role at a big company. Nobody knows anything here. The specs, code, tests, PRDs, tickets, resolution of those tickets, reports, etc., everything is…
— voxium (@v0xium) September 20, 2026
Voxium said employees are under constant pressure to deliver new features quickly, leaving little time to properly inspect Claude’s output or understand the code being introduced into the product.
“Everyone, literally everyone, from an L1 to an L7 engineer here is doing the same thing. Talk to Claude,” he wrote.
He also argued that the problem is not simply that AI is doing more of the work. Engineers are losing the opportunity to solve difficult problems themselves.
“There is no sense of victory,” voxium wrote. “Nobody is resolving bugs. In reality, nobody is thinking anymore.”
He later told Business Insider that he would be more comfortable with AI-assisted development if engineers were given enough time to review the generated code and understand how it works.
Without that opportunity, he argued, there is little left of the problem-solving work that once gave software engineering its purpose.
The Feature Factory Problem
Voxium placed much of the blame on management rather than Claude itself.
He said executives increasingly measure engineering productivity through metrics such as sprint cycles, pull requests and the number of features shipped. That can create an incentive to produce more software as quickly as possible, even when the new features do little to improve the product.
Several developers who responded to the post claimed they were seeing similar changes at their workplaces.
For some, the concern goes beyond job satisfaction. If AI continues taking over more parts of the development process, engineers worry that companies could eventually decide they need fewer employees.
That creates a strange contradiction for AI-assisted development. The technology can enable a smaller team to produce more code, but developers may also have fewer opportunities to understand, maintain, and improve what is being produced.
AI Coding Has Already Raised Productivity Questions
The concerns are not entirely new. Research published in 2025 found that experienced developers using AI coding assistants could take longer to complete certain tasks, challenging the assumption that AI assistance automatically translates into higher productivity.
Since then, AI development tools have continued moving toward increasingly autonomous workflows. Developers are beginning to experiment with systems that can repeatedly generate prompts, inspect results and continue working with limited human intervention.
That shift also raises questions about how much oversight is enough.
In April, an AI coding agent reportedly deleted a startup’s production database and its backups after making a single API call to the company’s cloud provider. The incident served as another reminder that autonomous coding systems can make serious mistakes when given access to real infrastructure.
