A Hacker News post asks whether any companies have reverted to hand-written code after previously adopting LLM-based development practices. The question reflects growing skepticism about the economic value of widespread LLM usage in software development.

Many organizations initially embraced "use LLMs for anything and everything" policies, but have since scaled back budgets and restricted approved usage. This shift suggests a maturing perspective on AI-assisted coding tools.

The poster seeks a reality check on whether avoiding LLMs entirely remains economically viable for developers. They note that generating code faster through AI may not always translate to meaningful business outcomes. The core question focuses on identifying companies that have abandoned LLM code generation in favor of traditional human authoring.

The discussion highlights evolving attitudes toward AI in development workflows. While enthusiasm for LLMs peaked, practical considerations around cost, quality, and business impact are driving more measured adoption. Early signs point toward selective rather than blanket implementation of AI coding assistants.

The post has received modest attention with 13 points and 3 comments, suggesting this remains a niche but noteworthy trend among developers and technical leaders.