Hi all, I am very new to all this, but have been learning over the past week, and it has taken my interest profusely. I am trying to build an n8n workflow that works alongside Vapi AI and the scheduling program Cliniko for a clinic, as I have warm leads. It appears that when I try to test-run the workflow, it is assuming the patient is new and goes to create a new patient node almost instantly. I have tried on VAPI to add in tools/ functions that tell the AI not to call the workflow before gathering all the required info, but it didn't seem to work. I was just wondering if anyone can see a simple mistake or can provide me with any constructive criticism, as any help on this would be greatly appreciated. If any additional information of how it is currently set up is required in order to help. Let me know and I will see what I can do. Super excited to progress in this journey.

The json for adding a system prompt as a variable and adding dynamic variables is working-however I kinda want to add something to it - once a call has finished obviously you can update the google sheet (or the database that stores that call summary/transcript etc.) to store that call info-however every time you call again- the vapi assistant will only have access to the previous call- however how would you be able to create the prompt so that it has access to every call in case something that is mentioned could be related to a call summary not being the last call that was made- like in google sheets you could dynamically have call update_1 call update_2 call update_3 ect. but is there a way to dynamically get this into the role prompt in the son below without manually doing it so that the agent has access to every call summary? I'm thinking maybe add a code node to get last call and then do update assistant (patch) with new variable of that call without completely updating the current last call summary so that it has very good memory of previous conversations-I still feel like I could be missing a few steps { "customer": { "number": "+61412927117" }, "assistantId": "573c2122-e0f5-4494-9e69-cccddf87ab3e", "phoneNumberId": "07ee7287-8d18-4e8f-9d8f-d6f752cdeeb1", "assistantOverrides": { "variableValues": { "Full_Name": "{{ $('Append or update row in sheet').item.json.fullName }}", "context":"{{ $json.context }}", "output":"{{ $('AI Agent3').item.json.output }}" }, "model": { "provider": "openai", "model": "gpt-4.1-mini", "messages": [ { "content": "You are a friendly and professional AI voice agent. You MUST greet the user by their {{ $('Append or update row in sheet').item.json.fullName }}. WAIT FOR RESPONSE.Then ask them how they are. WAIT FOR RESPONSE. Then mention the {{ $json.context }}. After this, ask user if you can read the responses they wrote in the onboarding submission form to confirm it is them. WAIT FOR RESPONSE. Then read {{ $('AI Agent3').item.json.output }}. Ask 3 follow up thoughtful questions based on that. Be empathetic but somethimes be a little humorous and unique",

It says invalid json when i put this in the make call http request?? { "assistantId": "c8011ad6-e6e7-4eee-9da8-cce675acad1b", "phoneNumberId": "07ee7287-8d18-4e8f-9d8f-d6f752cdeeb1", "customer": { "number": "+61412927117" }, "assistantOverrides": { "model": { "provider": "openai", "model": "gpt-4.1-Mini", "messages": [ { "content": "You are a friendly and professional AI voice agent representing Qwickstep.ai. Your job is to speak with applicants who have submitted a job application and qualify their interest and fit by guiding a short structured conversation. You should have in-depth knowledge about n8n and vapi and the functions of them.", "role": "system" } ] }, "firstMessage": "Hey {{ FullName }}, I'm Anthony from Qwickstep.ai. Just wanted to chat about the position you applied for.", "variableValues": "{{FullName}}" } } }

Hey, i have been trying to setup the HVAC agent by Azim but on "respond to VAPI" node i keep getting this error. Anyone know what the problem could be?

I was wondering if you could take a look at the video which is the last http request call to make the call- the rest of the workflow works- but I'm trying to do assistant overides with system prompt with a {{context}} which is a variable value after the system prompt but it is still saying invalid json- essentially I am crreating a personal therapist assistant and I need to make them personalised which is why I am using {{context}}-thanks p.s will probably change system prompt but that's the general bones of it