In this episode of An Average American, Sergey revisits the role of AI in modern software development after two years of hands-on experience. Early on, GitHub Copilot impressed but fell short of its lofty productivity claims. Today, with OpenAI Codecs and Cloud Code running in agent mode, the story is very different. Sergey reflects on how improved models like GPT-5 Codex and Claude 4.5, combined with better tooling, have realistically doubled his productivity on many tasks. He discusses why AI tools have finally matured, where they still lag — especially in speed — and how the developer workflow itself is evolving. This episode explores how AI has shifted from a novelty to an essential teammate in the coding process.