- The author ported Photoshop 1.0 (100k lines of Pascal and 20k lines of 68k assembly) to C# in 30 minutes using Claude Code, resulting in a functional but not perfect app.
- LLMs can handle less popular languages by understanding general software patterns, not just relying on training data volume.
- Coding agents make it easier to port prototypes from languages like Python to high-performance languages like Go or C# for scaling.
- Agents could replace cross-platform frameworks by porting native code between platforms (e.g., Swift to Kotlin) for best-of-both-worlds native apps.
- Agents can accelerate transitioning from memory-unsafe languages like C/C++ to Rust, improving security in legacy codebases.
- LLMs may enable new programming languages by allowing porting of existing library ecosystems, removing a major adoption barrier.