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

# Define how Aura behaves

> Set Aura's name, tone and model, its messages and instructions, and what it may do in a chat.

Everything under **Define** in the agent's menu shapes how Aura talks and what it's allowed to do. Changes go live when you click **Deploy changes** at the top of the page.

## Personality & tone

![The Personality and tone tab](https://mintcdn.com/emplorium/ugrIr0OwIbcsQUR_/images/aura/personality.png?fit=max\&auto=format\&n=ugrIr0OwIbcsQUR_\&q=85\&s=afc903f5cce4f94e53ff3cda51a8db9c)

Personality & tone: how Aura presents itself and writes.

| Setting | What it does |
| - | - |
| **Name & avatar** | The name and picture visitors see in the chat. |
| **AI model** | The model that powers Aura's answers. |
| **Escalation** | Lets visitors ask to talk to a person during an AI chat. |
| **Creativity** | Lower is more focused and consistent; higher is more varied. |
| **Tone** | The communication style of Aura's replies. |
| **Response style** | How brief or detailed replies are, and how much Aura uses emojis. |
| **Personality traits** | Traits that shape how Aura interacts. |

## Prompts

![The Prompts tab with instructions and messages](https://mintcdn.com/emplorium/ugrIr0OwIbcsQUR_/images/aura/prompts.png?fit=max\&auto=format\&n=ugrIr0OwIbcsQUR_\&q=85\&s=87bfecf966e7de1174082789f02c64f3)

Prompts: Aura's standing instructions and its fixed messages.

* **Instructions**: what Aura should always keep in mind, on top of its base instructions. For example: "Always answer in the visitor's language" or "Never promise delivery dates".
* **Messages** Aura sends word for word:
  * **First message**: the greeting visitors see when a chat starts.
  * **Fallback message**: what Aura says when it doesn't know the answer.
  * **Credit exhausted message**: what Aura sends if it runs out of [AI credits](https://docs.emplorium.io/billing/ai-credits.md) mid-chat.

## Tools

Turn on what Aura may do during a chat.

![The Tools tab with toggles](https://mintcdn.com/emplorium/ugrIr0OwIbcsQUR_/images/aura/tools.png?fit=max\&auto=format\&n=ugrIr0OwIbcsQUR_\&q=85\&s=bc28df0f0d0340781099a62194f2da17)

Tools: switch on what Aura is allowed to do.

| Tool | What Aura can do |
| - | - |
| **General knowledge** | Answer questions beyond your sources, from its general training. |
| **External data access** | Fetch data from the external APIs you've set up as [AI actions](#ai-actions). |
| **Transfer chat** | Hand the chat to a person on your team. |
| **Close chat** | Close the chat once it's resolved. |

Tools marked **Coming soon** can't be turned on yet. Some tools depend on your plan.

## AI actions

AI actions let Aura look things up in your own systems during a chat, like an order's status or a booking, and reply with real data.

![The AI actions tab](https://mintcdn.com/emplorium/ugrIr0OwIbcsQUR_/images/aura/ai-actions.png?fit=max\&auto=format\&n=ugrIr0OwIbcsQUR_\&q=85\&s=e747fd791c279ac6832dbff255bcbc79)

AI actions: ready-made actions, and actions connected to your systems.

* **Ready-made actions** come with Aura: turn on the ones it may use.
* **Your actions** connect to your own systems. Each one calls your API with a read-only `GET` request over HTTPS. You describe the inputs Aura must collect (an order number, an email) and the result it gets back, and you can **test** it with sample data before turning it on.
* **Context and history**: context fields are details Aura picks up in a chat, like an order number, that actions can reuse so visitors don't repeat themselves. The history logs every time Aura used an action, newest first, with sensitive inputs hidden.

> [!NOTE]
>
> AI actions depend on your plan. If they're locked, the tab shows **Unlock AI actions**.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.