AI Chat Assistant

Experimental

The AI assistant is experimental and under active development. Features, settings and behavior may change with any release.

nextCP/2 includes a built-in AI assistant that lets you control your audio setup with natural language. Open it via the AI Bot entry in the sidebar. You can ask the assistant to select a player or a music server, start a radio station, or switch the application language — the assistant performs these actions directly through nextCP/2.

Note

The AI assistant requires a configured AI provider (Google Gemini or any OpenAI-compatible endpoint). See AI Configuration for setup instructions.

What the assistant can do

The assistant has access to a set of tools inside nextCP/2. With these it can:

Capability	Example prompt
List and select media renderers (players)	“Which players are available?” / “Use the living room player.”
List and select media servers	”Show me my music servers.” / “Switch to the NAS server.”
List radio stations of the selected player	”What radio presets does my player have?”
Play a radio station	”Play Radio Paradise in the living room.”
Switch the UI language	”Switch the language to German.”

When the assistant selects a player or server, the selection is persisted and shown in the chat header (Player / Server), so you always know which devices the assistant is acting on.

Chat window

• Message input — type your request and press send. While the assistant works, an “AI is typing …” indicator is shown; the answer streams in as soon as it is available.
• New chat — clears the conversation history and starts a fresh session.
• History — the conversation is stored on the server and restored when you reopen the page, even from a different browser or device.

Conversation memory

By default each message is answered independently. If you enable conversation memory in the AI settings, the assistant receives the most recent messages of the current conversation as context (up to 20 completed messages), so follow-up questions like “and now the next station” work as expected.

Caution

Conversation memory increases token usage with every request. Keep it disabled if you use a paid, metered API and don’t need follow-up context.

Voice control via Alexa

The AI assistant can also be reached through an Amazon Alexa custom skill. The skill forwards the spoken text to the endpoint POST /api/voice/alexa, the assistant processes it with the same tools as the chat window, and the answer is read back by Alexa. Voice requests do not appear in the chat history of the web UI.