AI Producer & Roleplay Suggestions ​

When chatting or roleplaying with your companion, you might sometimes wonder: "What should I say next?" or "Where should this story go?"

AIRI includes a dedicated AI Producer Subsystem—a secondary intelligence loop that runs alongside your active conversation to suggest creative responses, dramatic plot twists, and topic changes in real time.

1. How the AI Producer Works ​

While your primary LLM generates character responses, the AI Producer analyzes the conversational flow in the background:

[ Conversation History ] ──► [ AI Producer Intelligence ]
                                       │
        ┌──────────────────────────────┴──────────────────────────────┐
        ▼                                                             ▼
 [ User Reply Suggestions ]                                 [ Director Scene Notes ]
 (3 quick-click response chips)                           (Background narrative direction)

• Zero Disruption: The Producer runs asynchronously without adding latency to character speech or replies.
• Context-Aware: Suggestions adapt to current mood states, active storyline scenarios, and recent Echo Chips.

2. Using Reply Suggestions in Chat ​

Above the message composer in the Desktop Chatbox, you will see 2–3 clickable Suggestion Chips:

• Playful / Banter: Quick witty replies or jokes.
• Inquisitive: Follow-up questions exploring what the companion just said.
• Narrative Action: Roleplay action prompts (e.g., *Hands her a cup of hot tea* or *Glances out the window at the rain*).

Triggering Suggestions ​

• Click to Send: Clicking any suggestion chip instantly sends that message into the conversation.
• Click to Edit: Shift-clicking (or clicking the edit icon on a chip) loads the suggestion into your text input box so you can customize it before sending.
• Refresh: Click the "Magic Wand / Reroll" icon in the composer toolbar to generate fresh suggestions on demand.

3. Configuring the Producer ​

You can customize the Producer's personality and creativity in Settings → Modules → Producer:

• Creativity Level: Adjust temperature to balance between natural grounded responses and wild narrative plot twists.
• Dedicated Producer Model: You can assign a fast, ultra-cheap model (such as Gemini Flash or a local Qwen 1.5B) to handle Producer suggestions without consuming your primary reasoning model's token limits.