A software developer detailed extensive daily use of large language models (LLMs) like Claude Code and Codex, spending hours interacting with AI-generated text across work and home environments, according to a July 8 blog post on alecscollon.com. The developer described a shift in their role from coding to designing code, instructing LLMs, reviewing AI-generated code, and then writing final versions.
The developer explained that their current work involves creating a framework for large-scale, unsupervised code generation within their codebase. They use Claude for tooling and sift through outputs from an unsupervised agent named Qwen. Despite not engaging with advanced autonomous agents or orchestration, the developer relies heavily on LLMs to explore unfamiliar coding approaches and gain comfort in less familiar areas.
This account highlights how LLMs are increasingly integrated into software development workflows, changing traditional coding roles. The developer’s experience reflects a broader trend where AI tools assist with ideation and code generation, potentially increasing productivity but also leading to what they describe as 'LLM burnout' due to constant interaction with AI content. The narrative offers insight into the evolving relationship between developers and AI assistants.
The blog post was published on July 8, 2026, providing a firsthand perspective on the impact of LLMs in everyday programming tasks and the challenges of sustained AI reliance, as detailed on alecscollon.com.