> For the complete documentation index, see [llms.txt](https://aizy.gitbook.io/aizy-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://aizy.gitbook.io/aizy-docs/agents-profile.md).

# Agent Profile

### 1. What you can configure

In **Agent's profile**, you define the identity of the selected bot. The section is organised in four tabs:

* **Autogeneration**: reads your public website and fills in the agent with what it finds; you review and adjust it afterwards.
* **Agent profile**: agent name and logo in a shared card, plus the agent description and purpose.
* **Privacy policy**: data-processing notice shown in the web chat.
* **Language & welcome message**: reference language, per-language behaviour checkboxes and the welcome message per language.

The section header shows the agent's id and, below it, the environment id (**Env. ID**), both useful when requesting support.

Localised messages can be edited in Spanish, English, Catalan, German, French, Italian, Dutch and Portuguese. The agent's behaviour and rules are not defined here but in **Conversational Prompt**.

### 2. Agent profile

The **Agent profile** tab lays out the agent's visible identity and its purpose in two columns:

* **Agent name** and **Agent logo** in PNG, JPG or WebP (max 100 MB), in a shared card. When you choose the file, the preview updates before saving.
* **Agent description & purpose**: explain in a few lines why the agent exists and what it should achieve. The agent receives this text as its purpose in every answer, both in the standard mode and with **Conversational Prompt**. The rules and flow of the conversation do not go here, but in those instructions.

### 3. Language & welcome message

The **Language & welcome message** tab brings the welcome message and the agent language together in two columns.

The **Welcome message** is edited per language with its selector and translation link.

The **Language** selector sets the agent's reference language. Below it, two checkboxes decide how it behaves during the conversation:

| Checkbox                                        | What it does                                                                                                                                                                                                |
| ----------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Always start in the selected language**       | The agent ignores the visitor's browser language and starts every new web chat conversation in the language selected here. It does not block an explicit language change requested during the conversation. |
| **Adapt to the language the visitor writes in** | The agent answers in the language each visitor uses and follows them if they switch during the conversation. Uncheck it to always answer in the language selected above.                                    |

Language adaptation is automatic and enabled by default. You do not need to configure any task to change language.

### 4. Privacy policy

The **Privacy policy** tab adds to the web chat the data-processing notice you decide on. It reaches the widget, the Chat Window and the installable app; it does not affect WhatsApp, Instagram or voice calls. On Telegram, **Informative only** mode sends the notice, with the policy URL, as its own message when each new conversation starts; **Blocking** mode does not apply there because Telegram cannot collect the acceptance.

| Field                                                | What it does                                                                                                                                                                                                                                                                   |
| ---------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| **Mode**                                             | **Disabled** shows nothing. **Informative only** adds the notice after the bot's first message and lets the visitor write. **Blocking: ask for acceptance** also adds the question with **I accept** and **I do not accept** and locks the composer until the visitor accepts. |
| **Privacy policy URL**                               | Shown as a link next to the notice. It must start with `https://`.                                                                                                                                                                                                             |
| **Notice text**                                      | Per language, with its selector and translation link. Leave it empty to use the default Aizy text.                                                                                                                                                                             |
| **Closing message when the visitor does not accept** | Only in blocking mode: if the visitor selects **I do not accept**, this message is shown and the conversation stops.                                                                                                                                                           |

Acceptance is remembered in the visitor's browser for that bot. A rejection is not stored: reloading the page asks again.

### 5. Autogenerate from your website

In the **Autogeneration** tab you type your website URL under **Autogenerate with your website** and select **Autogenerate**. Aizy reads the public website and applies and saves the following in the agent, without going through **Save**:

* Business description, which replaces the agent description and purpose.
* Conversation instructions, which replace the current ones.
* Visual colour palette.
* Widget welcome card message, in each language.
* Up to three initial buttons.

Because it is applied immediately, always review the result and correct it afterwards in the relevant tabs and sections. The detected palette is also available as **Autogenerated** in the widget settings. If the website is protected against automatic reading, Aizy says so on screen.

### 6. Save and check

Select **Save** in the section's bottom bar. The button saves at once every field across the **Agent profile**, **Privacy policy** and **Language & welcome message** tabs, whichever one is open. You can then use the agent preview to check identity, initial message and behaviour. Autogeneration does not replace human review of wording, colours or instructions.