AI model configuration

Volcengine Ark (recommended)

1. Sign in to the [Volcengine Ark console](https://console.volcengine.com/ark) and create an **inference endpoint**.

2. Copy the endpoint ID (format `ep-xxxxxxxx`).

3. Enter your API key under **AI Settings**.

4. When editing a bot:

- Provider: Volcengine

- Model: paste the `ep-` endpoint ID

**Common mistake**: using product names like `doubao-pro-32k` as the model ID will fail.

Other providers

ScalerDeep supports OpenAI-compatible APIs, DeepSeek, Zhipu GLM, Qwen, and more. Add keys in **AI Settings**, then select provider and model per bot.

Mock mode

Without API keys, the system uses Mock replies for UI testing only — not for validating real analytics or support quality.

Balance and availability

The AI Settings page shows whether each provider key is configured. The preview drawer also warns about Mock mode, missing keys, or invalid model IDs.