Get Call Messages

Retrieve the LLM conversation messages for a specific call. This returns the actual messages exchanged between the system, user, and AI assistant during the call - including system prompts, user inputs, and AI responses.

Messages vs Transcripts: Messages are the LLM conversation log (what the AI “thought” and said). Transcripts are the raw speech-to-text output of what was actually spoken. Use messages for debugging AI behavior; use transcripts for the actual conversation content.

Path Parameters

Parameter	Type	Required	Description
call_id	integer	Yes	The internal call ID (not the call SID)

Query Parameters

Parameter	Type	Required	Description
role	string	No	Filter by message role: system, user, or assistant

Request

curl "https://api.burki.dev/api/v1/calls/101/messages" \
  -H "Authorization: Bearer YOUR_API_KEY"

With Role Filter

curl "https://api.burki.dev/api/v1/calls/101/messages?role=assistant" \
  -H "Authorization: Bearer YOUR_API_KEY"

Response

Returns an array of chat message objects ordered by message index.

[
  {
    "id": 1,
    "call_id": 101,
    "role": "system",
    "content": "You are a friendly customer service assistant for Acme Corp...",
    "message_index": 0,
    "timestamp": "2024-01-15T10:00:00Z",
    "llm_provider": "openai",
    "llm_model": "gpt-4o-mini",
    "prompt_tokens": null,
    "completion_tokens": null,
    "total_tokens": null
  },
  {
    "id": 2,
    "call_id": 101,
    "role": "user",
    "content": "Hi, I'd like to check on my order status.",
    "message_index": 1,
    "timestamp": "2024-01-15T10:00:05Z",
    "llm_provider": null,
    "llm_model": null,
    "prompt_tokens": null,
    "completion_tokens": null,
    "total_tokens": null
  },
  {
    "id": 3,
    "call_id": 101,
    "role": "assistant",
    "content": "Of course! I'd be happy to help you check your order status. Could you please provide your order number or the email address associated with your account?",
    "message_index": 2,
    "timestamp": "2024-01-15T10:00:07Z",
    "llm_provider": "openai",
    "llm_model": "gpt-4o-mini",
    "prompt_tokens": 150,
    "completion_tokens": 35,
    "total_tokens": 185
  }
]

Response Fields

Field	Type	Description
id	integer	Unique message ID
call_id	integer	ID of the parent call
role	string	Message role: system, user, or assistant
content	string	The message content
message_index	integer	Position in the conversation (0-indexed)
timestamp	string	When the message was recorded (ISO 8601)
llm_provider	string	LLM provider used (e.g., openai, anthropic)
llm_model	string	Specific model used (e.g., gpt-4o-mini)
prompt_tokens	integer	Number of prompt tokens used
completion_tokens	integer	Number of completion tokens generated
total_tokens	integer	Total tokens for this message

Message Roles

Role	Description
system	System prompt and instructions for the AI
user	User’s spoken input (transcribed)
assistant	AI assistant’s response

Error Responses

404 Not Found

{
  "detail": "Call with ID 101 not found in your organization"
}

Use Cases

Debug AI Behavior

import requests

def analyze_call_conversation(call_id):
    response = requests.get(
        f"https://api.burki.dev/api/v1/calls/{call_id}/messages",
        headers={"Authorization": "Bearer YOUR_API_KEY"}
    )
    
    messages = response.json()
    
    # Get the system prompt
    system_prompt = next(
        (m["content"] for m in messages if m["role"] == "system"),
        None
    )
    print(f"System Prompt: {system_prompt[:100]}...")
    
    # Count exchanges
    user_messages = [m for m in messages if m["role"] == "user"]
    assistant_messages = [m for m in messages if m["role"] == "assistant"]
    
    print(f"User messages: {len(user_messages)}")
    print(f"Assistant messages: {len(assistant_messages)}")
    
    # Calculate token usage
    total_tokens = sum(
        m.get("total_tokens", 0) or 0 
        for m in messages
    )
    print(f"Total tokens used: {total_tokens}")

Export Conversation for Review

def export_conversation(call_id):
    response = requests.get(
        f"https://api.burki.dev/api/v1/calls/{call_id}/messages",
        headers={"Authorization": "Bearer YOUR_API_KEY"}
    )
    
    messages = response.json()
    
    # Format as readable conversation
    conversation = []
    for msg in messages:
        if msg["role"] == "system":
            continue  # Skip system prompt for readability
        
        speaker = "Customer" if msg["role"] == "user" else "AI"
        conversation.append(f"{speaker}: {msg['content']}")
    
    return "\n\n".join(conversation)

Analyze Token Costs

def calculate_call_llm_costs(call_id):
    response = requests.get(
        f"https://api.burki.dev/api/v1/calls/{call_id}/messages",
        headers={"Authorization": "Bearer YOUR_API_KEY"}
    )
    
    messages = response.json()
    
    total_prompt_tokens = 0
    total_completion_tokens = 0
    
    for msg in messages:
        if msg.get("prompt_tokens"):
            total_prompt_tokens += msg["prompt_tokens"]
        if msg.get("completion_tokens"):
            total_completion_tokens += msg["completion_tokens"]
    
    # GPT-4o-mini pricing (example)
    prompt_cost = total_prompt_tokens * 0.00015 / 1000
    completion_cost = total_completion_tokens * 0.0006 / 1000
    
    return {
        "prompt_tokens": total_prompt_tokens,
        "completion_tokens": total_completion_tokens,
        "estimated_cost": prompt_cost + completion_cost
    }

Notes

• Messages are stored in the order they occurred during the conversation
• The first message is typically the system prompt (role: system)
• Token counts are only available for assistant messages (LLM responses)
• For the actual spoken words, use the Get Transcripts endpoint