Typically I’d say “fork it”, but here, I would actually recommend designing your own language for its niche (a modern C, a performant language without Rust’s annoying memory management), having an LLM implement it, then open-sourcing with examples and measurements (how easy is it to work with? How many bugs are in generated code? etc.). These languages are bound to be so similar that, if one proves better than others, LLMs can port the others’ code. And it’s very fun, and you get a language with features specific to your use-case, and bike-shedding in your favor.