I use AI (Oct 2026) (ben.stolovitz.com)

In a recent reflection on his evolving use of AI, a research engineer at Salesforce AI Research shares how significantly his interaction with AI, particularly large language models (LLMs) and coding agents, has transformed over the past year. Moving from Microsoft, he's now integrating LLMs deeply into his workflow, spending approximately $3,000 a month on tokens. The most remarkable development has been the rise of "agentic coding," where agents like Claude autonomously perform tasks such as coding, testing, and bug fixing, fundamentally changing how software is built. This shift allows for rapid prototyping and testing of ideas, with the engineer noting that trying innovative solutions has become easier, although challenges around code quality and complexity remain. The significance of this shift for the AI/ML community is profound. It highlights a transition to a collaborative model between humans and AI, where agents bypass traditional coding drudgery and allow engineers to focus on higher-level design and strategic thinking. However, the engineer also raises concerns about the diminishing joy of programming and the psychological fatigue of managing multiple AI agents concurrently. As generative AI tools like Google’s AI summaries and sophisticated LLMs redefine search and research tasks, they offer new opportunities for learning and exploration, emphasizing the need for clear personal AI policies to navigate this rapidly advancing landscape.