+ 1 , agree still prefer composable pipeline architecture for voice agents. The flexibility on switching LLM for cost optimisation or quality is great for scal

+ 1 , agree still prefer composable pipeline architecture for voice agents. The flexibility on switching LLM for cost optimisation or quality is great for scaled use cases.