assistant message
**Assistant message** is the **response turn from the AI in a chat conversation** — the output generated by the model in response to user and system messages, forming the core interaction in conversational AI systems.
**What Is an Assistant Message?**
- **Role**: The AI's response in chat-based APIs.
- **Format**: {"role": "assistant", "content": "..."}.
- **Context**: Part of system → user → assistant message sequence.
- **APIs**: OpenAI Chat, Claude, Llama, all chat-format models.
- **Purpose**: Contains the model's generated response.
**Why Assistant Messages Matter**
- **Conversation History**: Include in context for multi-turn dialogue.
- **Few-Shot Examples**: Pre-fill assistant messages to demonstrate format.
- **Continuations**: Prefill partial assistant message for controlled output.
- **Format Control**: Show expected response structure through examples.
**Message Structure**
```python
messages = [
{"role": "system", "content": "You are helpful..."},
{"role": "user", "content": "What is Python?"},
{"role": "assistant", "content": "Python is..."}, # Previous turn
{"role": "user", "content": "How do I install it?"} # Current
]
```
**Prefilling Technique**
Start assistant message to control output format:
```python
{"role": "assistant", "content": "```json
{"} # Forces JSON output
```
Assistant messages enable **multi-turn conversations and format control** — core to chat-based AI.