This is a clever approach to reduce token usage. In my experience with Gemini 3 for code analysis, the biggest bottleneck isn't just the logic, but the verbosit

This is a clever approach to reduce token usage. In my experience with Gemini 3 for code analysis, the biggest bottleneck isn't just the logic, but the verbosity of standard languages consuming the context window. A targeted intermediate language like this could make 'thinking' models much more efficient for complex tasks.