The AI coding honeymoon is over, and companies are now measuring real impact. DolphinDB, a data platform company, shares its post-honeymoon reflections on AI-assisted development. They highlight Google CEO Sundar Pichai's revelation that 75% of new code is now AI-generated, up from 50% six months ago. Meanwhile, OpenAI engineers have largely shifted to their internal Codex tool, with about 99% of their output tokens coming from it. DolphinDB's takeaway is that AI coding is not a magic bullet but a tool that requires careful integration into workflows. The company emphasizes the importance of measuring productivity gains, not just token counts, and adapting engineering practices to leverage AI effectively. This reflection is valuable for engineering leaders who are navigating the transition from experimentation to production use of AI coding tools.
DolphinDB reflects on AI coding adoption after the initial hype, citing Google's 75% AI-generated code and OpenAI's internal Codex shift. The article focuses on realistic productivity gains and organizational learning.