Greetings from Puerto Rico! I’m new to the chat and, after trying to solve my question in the chat without success, I decided to come here to see if anyone can help. I want to integrate VAPI with langgraph.js to use it with a custom LLM. I’ve checked out the server message webhook documentation, but from what I understand, it only allows you to respond to specific events (like tool calls). My goal is simply to obtain the transcribed text, pass it through my LLM logic, and then return the text so VAPI can convert it to audio. I also looked into the custom LLM documentation, but I’m not entirely clear on how the formatted response works. From what I’ve gathered, you need to return the same response structure as OpenAI, yet in langgraph.js I’m only getting the text, and it seems odd that this is the only way to do it. Has anyone implemented something similar and could shed some light on this? Additionally, I’m exploring no-code solutions like Make.com, but I feel more comfortable working directly with code. Any suggestions? Thanks in advance!