I guess what I don't understand is that Vapi isn't really a front-end tech, it is just a way to interface LLMs, Voice models and transcribers, with the ability of creating assistants and logic to connect them. All of this can be managed via the API, so you don't need to use the UI at all. In fact I would even say that Vapi seems to have been designed in such a way that favors API usage, as the UI feels buggy in places still. Not sure if that helps