I use LLMs, December 2025 edition (hugues.betakappaphi.com)

A user in San Francisco reflects on their journey with large language models (LLMs) from initial frustrations to newfound productivity gains. Despite early setbacks—like ineffective assistance in design tasks and app development due to limitations in existing APIs—their experiences improved significantly by late 2025. Notably, they found significant value in using Claude, an LLM, for complex queries about corrective eye surgery, which helped them navigate options effectively and conduct thorough research that would have taken much longer on their own. This marked a turning point where the technology finally delivered on its promise of efficiency and quality. The implications of this story underscore a maturation in LLM capabilities, particularly in real-world applications like personal project assistance and technical troubleshooting. The user highlights advancements in command-line interfaces for coding with LLMs, facilitating safer and more efficient coding practices through version control and sandboxing. As LLMs become more adept at handling complex inquiries and integrating into users' workflows, they signal an essential shift in how individuals can leverage AI to enhance personal productivity, manage home projects, and handle technical tasks more effectively.