​
Inbound Calls - Voice Simulations

To simulate inbound calls, simply add your agent’s phone number in an evaluation template.

Requirements: Add your agent phone number in a Coval Template.

​
Outbound Calls

To simulate outbound calls, Coval requires an endpoint to trigger the outbound call with our agent’s phone number as payload.

Requirements: Provide us with an endpoint to trigger the outbound call (that takes in our number in the payload). This is to have the ability to trigger a call from your agent to our platform.

​
Text Simulations

If you have a chat agents or want to simulate your voice agent conversations without calls, you can use Coval the same way it’s used for voice simulations by generating text conversations (create test sets, metrics, templates)

Requirements: Provide us with a custom text endpoint

# Step 1: Get the chat or call ID
def get_chat_id(agent_id):
    url = "https://dev.api.examplecompany.ai/eval/get-call-id"
    payload = {"agentId": agent_id, "someExtraArg": extra_arg}
    try:
        response = requests.post(url, json=payload)
        response.raise_for_status()
        return response.json()
    except requests.exceptions.RequestException as e:
        print(f"An error occurred: {e}")
        return None

# Step 2: Send the call ID and messages to the custom LLM endpoint
def send_to_custom_llm(call_id, messages):
    url = "https://dev.api.examplecompany.ai/custom-llm/chat/completions"
    payload = {
        "call": {"id": call_id},
        "messages": messages,
        "customArgs": custom_args
    }
    try:
        response = requests.post(url, json=payload)
        response.raise_for_status()
        return process_sse_response(response.text)
    except requests.exceptions.RequestException as e:
        print(f"Error making request to custom LLM: {e}")
        return None

# Open AI like message format
messages = [
    {"role": "user", "content": "Hi, this is John Smith."},
    ...
]

# Completion call
result = send_to_custom_llm(call_id, messages)
print("Response from turn:")

# Returns same open AI message like format
print(json.dumps(result, indent=2))

Was this page helpful?