Would anyone be interested If I made sort of a "Voice AI SDK" that basically an abstraction layer that allows you to use differnt providers like Vapi, retell and other ones similar to AI SDK?