Handwriting code is the new cursive. AI agents write code competently, and they're improving fast. My recent agentic work spans refactoring C++ machine learning libraries, writing CLIs in Rust, building web apps in ASP.NET, and shipping mobile apps in Flutter. For the mechanical parts — scaffolding, boilerplate, repetitive transformations — agents handle it well. I've used them to write entire features on their own as well. It makes you wonder, if the agent writes the code, is the language is an implementation detail — and the intuition makes sense: why care about syntax you'll never type? But the language isn't just what the agent writes in. It's what the compiler checks, what the human reviews, and what determines how fast the feedback loop closes. Agentic coding raises the stakes for language design, and points toward specific properties a language should have. Better Compiler Feedback Multiplies Agent Productivity My agentic coding experience has varied by language.…