I like the concept we used at GitLab since beginning: boring solutions. Whenever you have to pick, you pick the boring one. It almost always pays off.