A Salesforce research engineer details how agentic AI has transformed his workflow over the past year, shifting from manual coding to managing multiple autonomous AI agents for development.
Key Points
- The author now uses agentic tools like Claude Code to handle over 90% of his programming tasks, including feature prototyping, bug fixes, and test creation.
- Managing multiple AI agents simultaneously has introduced new challenges, such as cognitive fatigue and the need for human oversight to prevent "over-architected" or low-quality output.
- Generative AI has become the default for research and search, though the author remains skeptical of its utility for creative writing and commercial art generation.
- The author has implemented a personal AI policy to establish ethical boundaries and maintain human agency as AI tools become increasingly integrated into daily life.
- Despite advancements in AI capabilities, the author emphasizes that human expertise remains essential for high-level architectural design and nuanced, original creative work.