> ## Documentation Index
> Fetch the complete documentation index at: https://docs.galtea.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Create WebRTC Connection

> Create a new WebRTC connection for a product.

## Returns

Returns a [WebRTC Connection](/concepts/product/webrtc-connection) object for the given parameters, or `None` if an error occurs.

## Example

```python theme={"system"}
web_rtc_connection = galtea.web_rtc_connections.create(
    product_id=product_id,
    name="production-voice-agent-" + run_identifier,
    api_key="YOUR_PIPECAT_CLOUD_API_KEY",
    agent_name="my-pipecat-agent",
    agent_speaks_first=False,
)
```

## Parameters

<ResponseField name="product_id" type="string" required>
  ID of the product this connection belongs to.
</ResponseField>

<ResponseField name="name" type="string" required>
  Name of the WebRTC connection. Must be unique per product.
</ResponseField>

<ResponseField name="api_key" type="string" required>
  The Pipecat Cloud **public API key** (`pk_...`) Galtea uses to reach your agent, the credential its runtime `/start` call needs. This is **not** the `pcc_pat_` Personal Access Token used to deploy the agent. This is a secret. Galtea stores it encrypted and the API never returns it. A fetched connection exposes only a `has_api_key` boolean, never the key itself.
</ResponseField>

<ResponseField name="agent_name" type="string" required>
  The name of your deployed Pipecat Cloud agent (for example, the `agent_name` you deployed with). It must match that agent's name exactly. Galtea passes it to Pipecat Cloud's `/start` call to launch the session; if no deployed agent has this name, the call fails.
</ResponseField>

<ResponseField name="agent_speaks_first" type="boolean" optional>
  Whether the voice agent opens the conversation once the session connects. Defaults to `False` (the simulated caller speaks first). Pass `True` for interviewer-style agents that greet the caller first.
</ResponseField>

<Note>
  The simulated caller's spoken language is taken from each test case's language, not the WebRTC connection, so one connection can serve test cases in different languages. See [WebRTC Connection](/concepts/product/webrtc-connection#how-the-caller-s-language-is-chosen).
</Note>