messages

model

messages

model

role: "system"

content

json
{
  "model": {
    "provider": "openai",
    "model": "gpt-4",
    "messages": [
      {
        "role": "system",
        "content": "You're a pizza ordering assistant. The user will ask for toppings, you'll add them. When they're done, you'll redirect them to checkout."
      }
    ]
  }
}

systemPrompt

/assistant/<ASSISTANT_ID>

curl --location --request PATCH 'https://api.vapi.ai/assistant/<ASSISTANT_ID>' \
--header 'Content-Type: text/plain' \
--header 'Authorization: Bearer <YOUR_API_KEY>' \
--data '{
  "model": {
    "provider": "openai",
    "model": "gpt-4o",
    "messages": [
      {
        "content": "You are a smart assistant who responds to user queries using the information you know, or information supplied by outside context.",
        "role": "system"
      }
    ]
  }
}'

POST https://api.vapi.ai/assistant

{
  "name": "Your Assistant Name",
  "model": {
    "provider": "openai",
    "model": "gpt-4o",
    "temperature": 0.7,
    "messages": [
      {
        "role": "system",
        "content": "Your system prompt goes here. This will define how the assistant behaves."
      }
    ]
  },
  "voice": {
    "provider": "playht",
    "voiceId": "jennifer"
  },
  "firstMessage": "Hello! How can I help you today?",
  "firstMessageMode": "assistant-speaks-first"
}