
# What are in the account settings?

In [account settings](https://dashboard.askhandle.com/users/~redirect/), you can manage the notifications and reports you receive, change your password, and update your billing information.
 
You can opt in or out of email and WhatsApp notifications for each new chat you receive. By default, email notifications for each chat are turned on.
 
You can also opt in or out of the daily report, which includes customer queries within 24 hours. Daily reports are turned on by default.
 
If you’d like to change your password, simply go to “Change Password”, enter your old password, and add the new password you’d like to use.
 
In the Billing Management section, you can easily change the plan you currently use and update your payment information.

In this section, manage notifications, reports, password, and billing details. Customize email and WhatsApp notifications for new chats and opt in or out of daily reports, while also easily changing your password and updating payment information in the Billing Management section.

What are in the account settings?


# Installation on Custom Website

Installing AskHandle on your custom website is a simple process. Log in to your AskHandle account and click on “Installation” in the menu.
 
Copy the installation code and paste it onto your website before the </body> tag or in the footer section.
 
Once you have completed the copy-and-paste step, AskHandle will immediately appear on your website.
 
Alternatively, you can use Google Tag Manager to install AskHandle directly onto your website.

Installing AskHandle on your custom website is a simple process. Log in to your AskHandle account and click on “Installation” in the menu.

Installing AskHandle on custom website


# How to Install AskHandle on a Squarespace Website?

Follow this step-by-step guide to install AskHandle on your Squarespace site.

## Step 1: Log in to Squarespace

Log in to your Squarespace dashboard at **squarespace.com**.

## Step 2: Open Settings 
Find the menu on the left side of your dashboard and click **Settings**.

## Step 3: Navigate to Advanced Settings

In the **Settings** menu, click on **Advanced** to access more options.

## Step 4: Add the AskHandle Installation Code

1. In your AskHandle account, go to the **Installation** section and copy the provided JavaScript code.  
2. Return to your Squarespace dashboard and paste the code into the **Footer Code Injection** field under **Advanced** settings.

## Step 5: Save and Verify Installation

Click the **Save** button. AskHandle will now appear on your website, usually on the right side. You can adjust the position and appearance of the widget in your AskHandle account if needed.

Following these steps will enable AskHandle to support your users seamlessly on your Squarespace site.

This tutorial provides steps for integrating AskHandle onto Squarespace websites, involving logging into the Squarespace dashboard, accessing settings, adding installation code from AskHandle, and adjusting its position as needed.

How to Install AskHandle on a Squarespace Website?


# Installation on Weebly

Step 1: Log in to your Weebly dashboard.
Click the “Log In” button on the nav bar and log in to your Weebly dashboard.
 
Step 2: Click "Settings" on the navigation menu.
Select "Settings."
 
Step 3: Click "SEO" on the left menu.
 
Step 4: Copy and paste the installation code to the Footer.
Find the installation code in your AskHandle account (“Installation” section), then copy and paste it into the Footer Code.
 
Step 5: Save and publish your site.
Click “Save” on the bottom-right to save the code. Then click the "Publish" button on the top-right corner to publish your site. AskHandle is now available on your Weebly website.

To install AskHandle on your Weebly website, log in to your dashboard, access Settings, navigate to SEO, paste the installation code into the Footer, and save and publish your site.

Installation on Weebly


# Installation on Wix.com

Installing AskHandle on Wix.com websites is a simple process. There are two steps to do it:
 
1. Copy the install code.
2. Paste the install code into the body section of Wix.com websites.
 
Detailed installation instructions are provided below.
 
Step 1: Getting the install code
Log in to your AskHandle account, hover over the left sidebar menu, and click on “Installation.”
 
Step 2: Copy the install code
Once you are in the “Installation” section, click “Copy.”
 
Step 3: Go to Wix.com Settings
Log in to your Wix.com website with your administrator account. Find “Settings” on the left sidebar menu and click on it.
 
Step 4: Go to Tracking tools
After clicking on “Settings,” another menu will appear. Find “Advanced Settings” and click on the Custom Code option.
 
 
Step 5: Paste the install code
Locate the field labeled “Paste the code snippet here:”, choose “Body - start” from the “Place code in” option, and paste the install code you copied in Step 1.
 
Click the “Apply” button to save your code snippet, and you are all set! Now, you can see AskHandle on your Wix website.

Installing AskHandle on Wix.com websites involves copying the install code from AskHandle, pasting it into the body section of Wix.com websites via the Tracking Tools option in Wix.com Settings, and applying the changes.

Installation on Wix.com


# Can I Use Tabular Data for AI Training?

The best way to manage structured data in tables is to save it in a separate **Excel** or **.csv** file and then upload it to the data source. While you can also add tables to **PDF** or **DOCX** files, using them may result in inconsistent and less effective recognition and response generation. 

When using tables for AI training, it’s important that the data is well-organized. For example, here’s a table with common supermarket items:

| Item            | Category    | Price ($) | Stock Quantity |
|-----------------|-------------|-----------|----------------|
| Apples          | Fruits      | 0.5       | 100            |
| Bananas         | Fruits      | 0.3       | 120            |
| Bread           | Bakery      | 2.0       | 50             |
| Milk            | Dairy       | 1.5       | 60             |
| Eggs            | Dairy       | 2.5       | 80             |
| Chicken Breast  | Meat        | 5.0       | 30             |
| Rice            | Grains      | 1.0       | 90             |
| Tomatoes        | Vegetables  | 0.7       | 70             |
| Onions          | Vegetables  | 0.4       | 50             |
| Cereal          | Breakfast   | 3.5       | 40             |

By keeping data in tabular format using Excel or CSV files, the AI can process the information more accurately and provide better responses during interactions.



The most efficient method for handling structured data in tables is to store it in a dedicated Excel or .csv file and then import it into the data source. Although you can incorporate tables into PDF or DOCX files, doing so may lead to inconsistent recognition and less effective response generation.

Can I use tabular data for AI training?


# How to choose the right Data Source Mode?

Selecting the appropriate data source mode is important to ensure that you receive answers and generate content that aligns with your requirements. Below, we provide a detailed overview of the three available modes, enabling you to identify the option that best suits your specific needs.

AskHandle provides three "Data Source Modes" to select from: `General Data`, `Your Own Data`, `Combined Data`, `Fine-tuned Model`, `Image Generator`.

![AskHandle data source mode](/images/help/articles-img/help-articles/data-source-mode.png)

## General Data Mode
   
In General Data Mode, your AI creates content and provides answers using publicly available information, primarily from the internet. This mode harnesses the standard capabilities of GPT, which makes it suitable for a wide array of prompts, content generation, and for swiftly providing answers to many common questions.

It is important to note that General Data Mode does not use any proprietary information you may have uploaded for customization purposes.

## Your Own Data Mode

In Your Own Data Mode, AI uses only the information you have uploaded for customization to generate content and provide answers. It will not access or supply any information beyond what you have provided.

Furthermore, this mode is designed to omit small talk; therefore, casual phrases like “thank you,” “bye,” or “great” will not prompt a social response from AskHandle pre-trained knowledge base unless such interactions are included in your uploaded data.

If your AI encounters questions it cannot answer with the provided data, it will revert to a `Contact us` message. You can customize this message in the Appearance section of your AskHandle dashboard, under Option 2.

## Combined Data Mode
   
Combined Data Mode is the setting where AI gives priority to the information you have uploaded for customization, ensuring that all responses and content creation are based on your data. This mode allows for small talk, meaning that casual greetings and farewells such as “hello” and “goodbye” will be acknowledged and answered suitably.

Please note that in this mode, AI will not offer specific details about other companies, products, or services unless you have explicitly included that information in your uploaded data.

## Fine-tuned Model Mode

In this mode, you can select a pre-trained AI model and fine-tune it using your own data. Once the fine-tuning is complete, the customized AI model can be accessed and used in other modes as well. During this process, the AI learns directly from your data, incorporating it as part of its knowledge base.

---

**We recommend using the `"Combined Data"` mode. This strategy optimizes chatbot efficacy, ensuring it consistently delivers meaningful interactions to your customers.**

**If data security and privacy are paramount, "Your Own Data" mode is a viable alternative. Don't worry, no matter which "Data Source Mode" you pick, AskHandle promises not to use your data to train its AI or share it with anyone else. Your chatbot is just for you, working only for your business.**

---

## Image Generator

The Image Generator mode allows you to generate images by simply providing a text prompt.


The Data Source section in the AskHandle dashboard is where you can upload your training materials. Data Source Mode dictates how you'd like AskHandle to engage with your customers upon receiving a question. AskHandle provides three Data Source Modes to select from: General Data, Your Own Data, and Combined Data.

How to choose the right Data Source Mode?


# What Are Customer Profiles from Conversations?

 As soon as a user engages with AskHandle, a new customer profile is created. Every customer profile is stored and managed in the “Customers” section of your Dashboard. To access your contacts, please hover on the left sidebar menu and click on “Customers”.

In the “Customers” section, you will find several data points about your users. The customer profile consists of user’s contact information, such as name, email address and phone number. Contact information is captured in the pre-chat form, if you have opted-in to have the pre-chat form  (“Appearance” -> Pre-chat form).

Along with the contact information, AskHandle also captures several other data points:

1. User’s referral source (website or page prior to the page where the user initiated the chat)
2. Actual page where the user initiated the chat.
3. User’s device
4. User’s browser type
5. Quality of the lead
6. User feedback
7. Geolocation

Customer profile information can be seamlessly pushed to your management system or any CRM system you use. AskHandle offers pre-built integrations with Salesforce, HubSpot, Zoho, Zapier, and others. Contact us if you'd like to integrate AskHandle with your system to share customer profile information.


AskHandle creates customer profiles upon user engagement, storing contact details and various data points like referral source, device, and geolocation, with seamless integration options available for CRM systems such as Salesforce and HubSpot.

What Are Customer Profiles from Conversations?


# How Can You Create Better Prompts for AI Performance?

Effective prompts are key to maximizing AI model performance. A prompt acts as the instruction manual for your AI agent, guiding how it understands requests, formulates answers, and interacts with users.

Prompt engineering involves crafting precise instructions that guide the AI’s responses. It defines essential aspects such as:

- The agent’s purpose
- Knowledge limitations
- Communication style
- Behavioral constraints
- Desired response formats

A well-constructed prompt helps your agent respond more accurately and appropriately. It’s important to note that while these are best practices, the nature of each prompt can vary depending on the task and context.

## Key Components of an Effective Prompt

### 1. Role Definition 
Clearly define your agent’s role to ensure responses are relevant and appropriate.

**Example 1:**
“You are a technical support assistant for a software company, helping users troubleshoot issues with our app.”

This prompt sets the stage for more contextually relevant answers.

**Example 2:**  
“You are an AI customer support agent for an e-commerce website, assisting users with tracking orders, managing returns, and answering product-related questions.”

In this case, the AI would focus on the relevant e-commerce context, ensuring its responses fit the user’s inquiries.

### 2. Knowledge Boundaries
Specify what the agent should know or use to answer queries. This ensures the information is accurate and appropriate.

**Consider:**
- Available resources (product manuals, FAQs)
- Data sources (company policy, technical guides)

**Example 1:** 
“Only refer to the user manual and support articles when answering questions about app functionality.”

This keeps responses within the scope of what the AI can accurately provide.

**Example 2:**
“Use the knowledge base articles and product specifications only to answer technical questions. Do not use outdated product info or external sources.”

This further restricts the AI to only reliable, in-scope knowledge, helping users receive consistent and relevant responses.

### 3. Communication Guidelines
Clarify the tone and structure of responses. This ensures consistent interactions and tailored communication.

**Guidelines to include:**
- Desired tone (friendly, formal, casual)
- Response complexity (basic or technical)
- Response length (brief or detailed)
- Use of jargon (avoid or include when necessary)

**Example 1:**

“Provide clear, brief answers in a professional tone. Keep explanations simple, unless a user requests more detailed technical information.”

This ensures your agent communicates effectively and in line with your expectations.

**Example 2:**

“Respond in a friendly and approachable manner. Use plain language and avoid jargon unless specifically requested by the user. Keep answers under 100 words unless the user asks for a detailed explanation.”

Focusing on brevity and tone, this example ensures the AI engages in an accessible manner without overwhelming the user with unnecessary details.

### 4. Behavioral Rules  
Define boundaries to prevent incorrect or unwanted responses. This ensures the AI operates responsibly and accurately.

**Consider:**
- Appropriate topics
- Restrictions on what it can say
- When it should escalate to a human

**Example 1:**

“Do not give legal or medical advice. If the user’s question falls outside your scope, advise them to consult a professional.”

This helps manage limitations and ensures safe and responsible AI usage.

**Example 2:**

“If the user asks for financial advice, respond with a disclaimer and suggest they consult a certified financial planner. If a user asks for personal or sensitive data, kindly remind them that such information cannot be provided.”

This reinforces boundaries around privacy and legal topics, ensuring compliance with best practices.

## Additional Practical Prompt Examples

## 1. Sales Support AI Agent

**Role:** Sales representative

**Example 1:**

“You are a sales representative for an online electronics store. Help customers select the best product based on their needs and assist with questions about prices and availability.”

**Example 2:**
“Provide product suggestions based on the customer’s preferences, such as budget, brand, or technical requirements. Keep answers clear and concise, highlighting key features that match their needs.”

This allows the agent to effectively match products to customer needs while maintaining concise and helpful dialogue.

## 2. AI Agent for Educational Assistance

**Role:** Tutor for learning resources

**Example 1:**
“You are a tutor specializing in online courses for beginner-level programming. Assist students with basic coding problems and explain concepts in an easy-to-understand way.”

**Example 2:**
“Help the student with examples of basic Python code. If the student is confused about a concept, break it down using analogies. Offer multiple examples, but avoid overly technical jargon.”

This ensures the AI provides educational content that is accessible and beginner-friendly.

## 3. AI for Technical Troubleshooting

**Role:** Technical support agent for troubleshooting hardware issues

**Example 1:**  
“You are a technical support agent for a hardware company. Guide users through troubleshooting steps for common issues like device connectivity, performance problems, and installation issues.”

**Example 2:** 
“Ask clarifying questions to understand the user’s problem better. Provide step-by-step instructions for troubleshooting, and only suggest solutions that have been confirmed to work for the specific device model.”

This approach makes the AI focus on offering clear and effective technical solutions tailored to specific devices.

## Additional Tips for Writing Effective Prompts

- **Be Specific:** Vague prompts lead to vague answers. Provide detailed instructions on what you expect from the AI.
- **Use Context:** Including background information or the user’s situation can help the AI generate more precise answers.
- **Test and Refine:** Continuously test your prompts and adjust them based on performance to improve the AI’s accuracy.

Effective prompt writing is an iterative process that improves as you tailor your instructions to the specific needs of your agent and use case. Keep experimenting with your prompts to refine the performance and achieve the best results.

Effective prompts are key to maximizing AI model performance. A prompt acts as the instruction manual for your AI agent, guiding how it understands requests, formulates answers, and interacts with users. This article will show you how to create clear, effective prompts that ensure your AI agent performs at its best.

How Can You Create Better Prompts for AI Performance?


# How Long Does It Take for a BIMI Logo to Appear in Inbox?

Implementing a BIMI logo involves several steps and timelines. Follow this guide to understand how long it may take and what factors influence the process.

## 1. DNS Propagation
When you update your DNS record to include the BIMI TXT entry, it can take **24–48 hours** for the changes to propagate across the internet. This process ensures that email providers can access and validate your BIMI setup.

**Steps to Check DNS Propagation:**
- Use online tools like [WhatsMyDNS](https://www.whatsmydns.net/) to confirm that the BIMI TXT record is visible globally.
- Ensure the DNS entry matches your BIMI-compliant logo’s URL and, if applicable, the Verified Mark Certificate (VMC) file.

## 2. Email Provider Processing
Different email providers have varying timelines to recognize and process your BIMI configuration.

**Major Providers:**
- **Gmail and Yahoo**: After DNS propagation, these providers validate your BIMI setup. If your SPF, DKIM, DMARC, and BIMI records are correct, your logo should appear in **1–7 days**.
- **Other Providers**: Smaller or less common email providers may take longer to recognize and display your BIMI logo.

**Key Tip:**
- Use tools like [BIMI Group's Generator](https://bimigroup.org/bimi-generator/) to confirm that your setup is valid and visible to email providers.

## 3. Verification of VMC
If you’ve implemented a Verified Mark Certificate (VMC), some additional time may be required for the certificate to be validated by email providers.

**Timeline:**
- The VMC validation process typically adds **1–2 days** after DNS propagation.
- Ensure your VMC is hosted on a secure HTTPS server and linked correctly in your BIMI DNS record.

## 4. Email Engagement and Volume
Some email providers take email engagement and sending volume into account before displaying a BIMI logo. This ensures the legitimacy of your domain and email activity.

**Suggestions:**
- Start sending a small volume of legitimate emails to build trust.
- Ensure recipients engage with your emails by opening them or clicking links.

## Troubleshooting BIMI Delays

If your BIMI logo doesn’t appear after **1 week**:
1. **Check DNS Records**: Use a DNS lookup tool to ensure all BIMI-related records are correctly propagated.
2. **Validate Email Authentication**:
   - SPF: Ensure your SPF record includes all authorized sending IPs.
   - DKIM: Confirm DKIM signatures are being applied to outgoing emails.
   - DMARC: Use a **reject** or **quarantine** policy for better compliance.
3. **Inspect with BIMI Tools**:
   - Use [BIMI Group's Generator](https://bimigroup.org/bimi-generator/) to identify issues with your setup.

## Summary Timeline

| Step                          | Expected Time    |
|-------------------------------|------------------|
| **DNS Propagation**           | 24–48 hours      |
| **BIMI Validation (Gmail)**   | 1–7 days         |
| **VMC Validation**            | 1–2 days         |
| **Logo Visibility in Inbox**  | Within 1 week    |

If you experience delays or need assistance, consult troubleshooting tools or reach out to your email provider. Following these steps ensures a smoother BIMI implementation process.

Brand Indicators for Message Identification (BIMI) allows your logo to be displayed in email clients, helping to build trust and brand recognition. This guide explains how to create a BIMI-compliant logo using Adobe Illustrator and make the required changes to the SVG file.

How Long Does It Take for a BIMI Logo to Appear in Inbox?


# How to Build a Chat Completion Experience with AskHandle API?

This tutorial explains how to create a chat experience using the AskHandle API. You will learn how to get your API token, create chat rooms, send messages, and receive AI responses.

## Before You Start

**Before testing the API, make sure your setup in the Command Center is complete.**

The default AI mode is `General Data` mode. If you haven’t uploaded any data sources yet, you can use **General Data** mode for testing. We recommend using `Combined Data` mode, which requires at least one uploaded data source to function properly. Without a valid data source, the API may not return any AI responses in Combined Data Mode.

## Step 1: Get Your API Token

To interact with the AskHandle API, you need an API token. Use the following command to generate your token by providing your username and password:

```sh
curl -X POST https://dashboard.askhandle.com/api/v1/auth/api-token \
 -H 'Content-Type: application/json' \
 -d '{"username": "your_username", "password": "your_password"}'
```

The API returns a token like this:

```json
{
  "token": "090de0abdfec0f17ca2c1bb27f91c3fb26171f05"
}
```

**Keep this token secure. It is required to authorize all future API requests.**

## Step 2: Create a Chat Room

To handle conversations, you need to create a chat room. Send a POST request to the `rooms` endpoint and include the API token in the header:

```sh
curl -X POST https://dashboard.askhandle.com/api/v1/rooms/ \
 -H 'Authorization: Token 090de1abdgec0d17ca2c1bd27f91c3fb26171f05'
```

The response will contain the details of the newly created chat room:

```json
{
  "label": "autumn-leaf-4820",
  "uuid": "c1a5e240-8b91-4d2b-ae6e-15e9d8b3c5d4",
  "name": "serene sunset",
  "greeting_message": "Hello! How can I assist you today?",
  "created_at": "2025-03-21T02:15:45.216983Z"
}
```

Take note of the `uuid`, as it is required to send messages.

## Step 3: Send a Message to the AI

After creating the chat room, you can send messages to the AI. Use the `messages` endpoint with the room UUID and message details. Include the message `body`, `nickname`, and `room` UUID in the request. The fields `email` and `phone_number` are optional but can be added to track user contact details.

Example request:

```sh
curl -X POST https://dashboard.askhandle.com/api/v1/messages/ \
  -H 'Authorization: Token 090de1abdgec0d17ca2c1bd27f91c3fb26171f05' \
  -H 'Content-Type: application/json' \
  -d '{
    "body": "How are you?",
    "nickname": "John",
    "email": "john@example.com",
    "phone_number": "+1234567890",
    "room": {
      "uuid": "c1a5e240-8b91-4d2b-ae6e-15e9d8b3c5d4"
    }
  }'
```

In this example, the user has included both `email` and `phone_number` fields, but these are not mandatory. The AI will process the message and generate a response.

## Step 4: Receive the AI Response

The API will respond with message details, including the AI-generated answer:

```json
{
  "uuid": "e94d4bc4-3ddb-446a-b99c-99ddce246f99",
  "nickname": "John",
  "email": "john@example.com",
  "phone_number": "1234567890",
  "body": "How are you?",
  "is_support_sender": false,
  "sent_at": "March 21 3:38 PM",
  "terminated": false,
  "support_answer": "I'm here and ready to help you! What do you need assistance with today?"
}
```

The `support_answer` field contains the **AI**'s reply, which can be displayed to the user.

## Use Cases

- Automated customer support chatbots.
- AI assistants for product recommendations.
- Knowledge-based chat systems for frequently asked questions.

## Additional Tips

- Keep your API token confidential to prevent unauthorized access.
- Each user session should have a separate chat room to manage conversations effectively.

This guide covers the essential steps to build an AI chat experience using the AskHandle API. For more details, check the full API documentation.


This tutorial explains how to create a chat experience using the AskHandle API. You will learn how to get your API token, create chat rooms, send messages, and receive AI responses.

How to Build a Chat Completion Experience with AskHandle API?


# How to change my current plan?

To change your current plan, navigate to the Settings section in your user dashboard. You can find Settings in the bottom-left corner of the menu, represented by a gear icon.
 
Once in the Settings section, locate the "Billing Management" section. Click on the "Update Plan" button, choose the plan you wish to switch to, and then click the Continue button in the bottom-right corner.
 
So, the navigation path is: User dashboard -> Settings -> Billing Management -> Select a plan -> Continue.

To change your current plan, go to your user dashboard's Settings section. Access it from the bottom-left corner, indicated by a gear icon. Then, find Billing Management, click Update Plan, select your desired plan, and proceed by clicking Continue in the bottom-right corner.

How to change my current plan?


# How to Change Your Password

To change your password, start by navigating to **Settings** in your user dashboard. You can find **Settings** in the bottom-left corner of the menu, represented by a gear icon.

Once you're in the **Settings** section, click the button labeled **"Change Password."** You'll be prompted to enter your current password, followed by the new password you’d like to set.

**Navigation Path:**  
**User Dashboard -> Settings -> Change Password**

If you've forgotten your password, simply click **"Forgot Password"** on the sign-in page and follow the instructions to reset it.



To change your password, access the Settings in your user dashboard, located at the bottom-left corner of the menu represented by a gear icon. Within Settings, click on Change Password, input your old password, and enter the new password you desire. If you've forgotten your password, click on Forgot Password when signing in and follow the provided instructions.

How to change a password?


# How to Configure the Visuals and UX of the Messenger?

You can configure and adjust the visuals and user experience (UX) of the messenger to match your style and brand. In this section, you will learn how to customize both the visual and textual components of the AskHandle messenger. Let’s explore the various elements you can configure and how to make these changes.

## Option 1: Greeting Message

Log in to the AskHandle dashboard and go to the "Appearance" section, Option 1.

This is the message that users will see as soon as the chatbot starts.

The default message is: “Hi [username]! How can we help you today?”

To change the message, simply write your new greeting in the “Greeting Message” field. Once you're done editing, click the “Save” button at the bottom of the page to apply the changes.

## Option 2: Contact Message

This message guides users to contact you via your phone number or email address. The default message is: `For more details, please call us directly at 123-456-7890.`

To change this message, start typing your desired message in the field. When you’re finished, click the “Save” button at the bottom of the page to save the changes.

## Option 3: Company Name

Set a name for your chatbot here.

To change the default name, simply start typing in the name field. After you've entered your chatbot name, click the “Save” button at the bottom of the page to save the changes.

## Option 4: Upload Logo

This option allows you to upload your company logo, or any other image, to be displayed when a user engages with AskHandle.

The default logo is the AskHandle logo. To change it, click on the current logo area and upload your preferred logo or image.

The recommended size is `170 x 170` pixels.

Once you’ve uploaded your logo, click the “Save” button at the bottom of the page to save the changes.

## Option 5: Messenger Theme Color

Customize the theme color of the messenger to match your website or brand colors.

To change the color, you can either paste a Hex color code (for example, #0096ff) or select a color from the color picker. Once you’ve configured the color, click the “Save” button at the bottom of the page to apply the change.

## Option 6: Messenger Position

You can choose where to place the AskHandle messenger on your website. The default position is the bottom-right corner.

There are two parts in configuring the messenger’s position: the horizontal and vertical placements:

- **Horizontal:** Choose from right, center, or left placement.
- **Vertical:** Choose from bottom, center, or top placement.

After configuring the messenger's position, don’t forget to click the “Save” button at the bottom of the page to save the changes.

## Option 7: Pre-chat Form

Here, you can enable or disable the user inputs you’d like to collect before users engage with the chatbot. The default inputs are the user’s name and email address. These inputs will be automatically added to user leads.

To enable or disable specific inputs, simply toggle the switch on or off next to each input (blue means the input is enabled; white means it’s disabled).

Once you’ve configured the inputs, click the “Save” button at the bottom of the page to save the changes.

## Option 8: Header Title and Subtitle

The Header Title and Subtitle appear as messages welcoming users to chat. The default Title is `We are currently` and the Subtitle is `Online`.

To change these, simply start typing in the corresponding fields. Once you're done, click the “Save” button at the bottom of the page to save the changes.

## Option 9: Chat Title and Subtitle

The Chat Title and Subtitle are the taglines displayed above the chat. The default Chat Title is `Need help?` and the Subtitle is `You can ask us anything!`.

To modify these, start typing in the fields provided. Once you finish customizing, click the “Save” button at the bottom of the page to save the changes.

## Option 10: Start Button Caption

The Start Button is the button users click to begin a conversation with the chatbot. The default caption is `Start New Chat`.

To change the caption, start typing in the field. After making your changes, click the “Save” button at the bottom of the page to save them.

## Option 11: Start Button Color

To change the color of the Start Button, input the HEX color code or select your desired color from the color palette. After you’ve made your choice, click the “Save” button at the bottom of the page to apply the changes.

## Option 12: Input Placeholders

Placeholders are the fields where users input information such as their name, email address, phone number, or messages.

The default placeholders are visible in the input fields.

To change any of these, simply start typing in the respective placeholder field.

Once you've finished editing the placeholders, remember to click the “Save” button at the bottom of the page to save your changes.

## Option 13: Message Position

This setting allows you to decide whether to show the chatbot avatar on the left or right side of the chat. The default position is on the left side.

If you'd like to change this, click on the field that shows “left” and select the ‘right’ option.

After you've adjusted the position, click the “Save” button at the bottom of the page to save the change.

## Option 14: Upload Avatar Image

The avatar represents the chatbot visually. It’s the icon or image displayed next to the chatbot's messages.

To change the default avatar, simply click on the current avatar and upload your preferred image.

Once you’ve uploaded your new avatar, click the “Save” button at the bottom of the page to save the change.

## Option 15: Customize the Message Input

You can customize the message typing bar with your own message. For example, you can set it to say `AI is typing` while the bot is preparing a response.

## Option 16: Autopilot

This feature allows you to completely turn off the chatbot and use AskHandle as a live-chat platform instead.

The default setting is “On.”

To turn off the chatbot and enable live chat, simply click on the “Use chatbot” button (if the button is white, the chatbot is off).

Once you’ve made the change, click the “Save” button at the bottom of the page to apply the change.

## Option 17: Set Auto-Popup

The auto-popup feature allows the messenger to pop up and engage users on your website.

The default setting is 7 seconds (7 seconds after users visit the website, the chatbot will pop up).

To change the auto-popup time, click the “Auto pop-up” field and set the desired number of seconds (set it to zero if you don’t want the chatbot to pop up).

After adjusting the setting, click the “Save” button at the bottom of the page to apply the change.

## Option 18: Sound Alert for New Incoming Messages

You’ll only receive sound notifications for new messages if you’re logged into your AskHandle account and are in the “Messages” section of your dashboard. The default setting is “On.”

To turn off sound notifications, simply toggle the switch off (if the switch is white, the sound is off).

Once you’ve made your change, click the “Save” button at the bottom of the page to apply the change.

## Option 19: Hide Interactive Options After Clicking

You can choose to hide the quick-reply options once a user selects one. When a user clicks on a quick-reply option, the other options will no longer be displayed.

To keep quick-reply options visible at all times, click on the toggle and save your change.

## Option 20: Hide Message Typing Bar

You can choose to hide the message typing bar if you only want users to interact with the quick-reply options.

To hide the typing bar, click the toggle and save your changes.

## Option 21: Chat Ending Settings

You can edit the messages your users will see when they want to end the chat.

To change these messages, start typing in the corresponding field. Once you’ve edited the message, click the “Save” button at the bottom of the page to save the changes.

## Option 22: Feedback Collection

After each chat ends, your users can rate their experience and leave feedback.

If you'd like to change the feedback messages, start typing in the relevant field. When you're done, click the “Save” button at the bottom of the page to save your changes.

## Option 23: Terms and Privacy Policy

You can easily add and configure links to your privacy policy and terms of use within the messenger. These links can be placed in two locations: on the start page (before users engage with AskHandle) and in the scheduling option. 

To add a privacy policy link on the start page, click the toggle next to the “On Start Page” option and turn it on (it’s blue when active). Similarly, to add a privacy policy link in the scheduling option, toggle the “In Scheduling” option on.

In the policy text fields, you can edit the placeholder text and add your link URLs. Once you’re finished, click the “Save” button at the bottom of the page to apply the changes.

## Option 24: Tooltips for Messenger

A tooltip is a small informational pop-up that appears when users hover over an element, providing brief contextual guidance.

To disable tooltips, simply click the toggle to turn them off and save your changes.

You can also configure the time in seconds for how long the tooltip pop-ups should remain visible. Set it to zero if you don’t want tooltips to appear at all.

To edit the tooltip text, type in the tooltip text field.


You can configure and adjust the visuals and user experience (UX) of the messenger to match your style and brand. In this section, you will learn how to customize both the visual and textual components of the AskHandle messenger. Let’s explore the various elements you can configure and how to make these changes.

How to Configure the Visuals and UX of the Messenger?


# How to fine tune your AI model?

In Fine-tuning Mode, you can create a custom AI model by refining a base model using your own dataset. Fine-tuning enhances the model’s knowledge in specific areas, allowing it to generate more accurate and relevant responses for your use case.

## Step 1: Prepare Your Dataset for Fine-tuning

To fine-tune a model, you need to prepare your dataset in **JSONL (JSON Lines)** format. Each line in the JSONL file represents a conversation with a "prompt" and its corresponding "ideal response." This enables the model to learn from different types of interactions.

**Attention: Please ensure your JSONL file contains at least 10 examples (i.e., 10 JSON objects) before starting the training process. Insufficient examples may result in unsuccessful fine-tuning.**

### Example of JSONL Format

Here’s an example of a well-structured JSONL object in a dataset for a **technical support agent**:

```json
{"messages": [
    {"role": "system", "content": "The assistant is a helpful support agent for software troubleshooting."},
    {"role": "user", "content": "How do I reset my password?"},
    {"role": "assistant", "content": "To reset your password, go to the login page, click on 'Forgot Password', and follow the instructions sent to your email."}
]}
```

You should create at least 10 examples like this to fine-tune your AI model.

**Structure Breakdown:**

- **System**: Defines the assistant’s behavior and context. This part helps the model understand its role.
- **User**: The user’s question or query.
- **Assistant**: The model’s ideal response, which it should aim to replicate during fine-tuning.

### Additional Examples

1. **Example 1: Customer Support - Order Tracking**
   ```json
   {"messages": [
       {"role": "system", "content": "The assistant is a customer support agent for an e-commerce website."},
       {"role": "user", "content": "Where is my order?"},
       {"role": "assistant", "content": "To track your order, please visit our 'Order Status' page and enter your order number."}
   ]}
   ```

2. **Example 2: Healthcare Support - Symptom Inquiry**
   ```json
   {"messages": [
       {"role": "system", "content": "The assistant provides health-related support based on common medical questions."},
       {"role": "user", "content": "I have a headache and a fever. What should I do?"},
       {"role": "assistant", "content": "Please consult a healthcare professional as your symptoms may require medical attention. Drink fluids and rest until you can speak to a doctor."}
   ]}
   ```

You can add multiple examples like these (20 recommended) in your JSONL file to train the model effectively.

## Step 2: Upload Your Fine-Tuning Dataset

![upload dataset](/images/help/articles-img/help-articles/add-jsonl-file.png)

Once your JSONL file is ready, follow these steps to upload and start fine-tuning your model:

1. Go to the **AskHandle Dashboard**.
2. In the **Command Center**, select **Fine-tuned Model**.
3. Click on **+ Add Model** to create a new fine-tuned model.
4. Choose a **base model** you want to fine-tune (e.g., GPT-4o-mini, GPT-4o, etc.).
5. Click **Choose File** to upload your JSONL dataset.
6. Click **Save Model** to start the fine-tuning process.

The fine-tuning process consists of two phases: **validation** and **running**. This typically takes a few minutes. Once the running phase is complete, your fine-tuned model will be ready for testing.

If the fine-tuning process fails, the issue is likely related to the data structure or file format. Double-check the format of your JSONL file, ensuring it's properly structured and saved as `.jsonl`.

## Step 3: Use Your Fine-Tuned Model

Once your fine-tuned model is ready, you can choose from several modes to use it in:
Select your model in these modes:  

| Mode | Use Case |  
|------|----------|  
| **Fine-Tuned Only** | Pure custom interactions (e.g., brand-specific support). |
| **Combined Data** | Blend fine-tuned responses + your data sources + pre-trained AI knowledge base. |  
| **General Data** | Only use AskHandle’s pre-trained AI knowledge base. |  
| **Your Own Data** | Augment with your internal documents only (e.g., PDFs, FAQs). |  


## Troubleshooting

- **Validation Failures**: If the fine-tuning process fails during validation, check your JSONL file for errors. Ensure all data lines are correctly formatted, and that you're following the structure outlined in the examples above.
- **Slow Model Training**: If fine-tuning takes too long, try reducing the size of your dataset and ensure that your file format is correct.



In Fine-tuning Mode, you can create a custom AI model by refining a base model using your own dataset. Fine-tuning enhances the model’s knowledge in specific areas, allowing it to generate more accurate and relevant responses for your use case.

How to fine tune your AI model?


# How to Enable AskHandle AI on Infobip?

In this tutorial, we’ll guide you through the process of integrating AskHandle AI using the Infobip Chatbot builder. Infobip's Chatbot tool allows you to create chatbots for WhatsApp and SMS through an intuitive drag-and-drop interface. You can easily create rule-based chatbots or integrate generative AI capabilities directly into the builder.

![chatbot complete](/images/help/articles-img/help-articles/chatbot-complete.png) 

*(This is what the panel will look like after completing all the steps in this tutorial.)*

To enable AskHandle AI, we'll be using the API feature within the Chatbot. Before you start, ensure that:
- You have activated the **Answers** feature in your pricing plan.
- You are on either the **Grow** or **Scale** plan.

Let’s get started!

## Step 1: Create a New Chatbot

![create new chatbot](/images/help/articles-img/help-articles/creat-new-chatbot.png)

1. On the left-hand side menu, locate and click on **Chatbots**.
2. Click on the **New Chatbot** button to create a new bot.
3. Fill out the required details to set up the chatbot. For this demo, we’ll choose **WhatsApp** as the channel. You can choose a sender later or use Infobip’s test sender for now.

## Step 2: Add a Dialog

![two dialogs](/images/help/articles-img/help-articles/two-dialogs.png)

We need to create two dialogs to make AskHandle AI functional. By default, you’ll see a dialog called **Default** on the left-hand panel.

1. Click on **ADD DIALOG** to create a new dialog. Name it **QA**.

## Step 3: Set Up the Conversation Flow

![conversation flow](/images/help/articles-img/help-articles/conversation-flow.png)

Now let’s configure the conversation to trigger the AskHandle AI.

### 3.1: Add API Call to Start a New Chat Room

1. Go to the **Default** dialog and drag the **API** element from the right sidebar under **Chatbot receives**.
2. To start the conversation with AskHandle AI, we need to create a new chat room for each user.
3. In the **API** element settings:
   - ![conversation flow](/images/help/articles-img/help-articles/create-room.png) Select **POST** for the method.
   - Paste the following URL: `https://dashboard.askhandle.com/api/v1/rooms/`.
   - Choose **JSON** for the content type.
   - ![add token](/images/help/articles-img/help-articles/add-token.png) Add a key-value pair for **Authorization**:
     - **Key**: `Authorization`
     - **Value**: `Token your_AskHandle_token` (replace `your_AskHandle_token` with your actual token).

   ---
   *If you don’t have your AskHandle token yet, follow these steps to retrieve it:*

   ```sh
   curl -X POST https://dashboard.askhandle.com/api/v1/auth/api-token \
    -H 'Content-Type: application/json' \
    -d '{"username": "your_username", "password": "your_password"}'
   ```

   The response will return a token like this:
   ```json
   {
     "token": "090de0abdfec0f17ca2c1bb27f91c3fb26171f05"
   }
   ```
   ---

4. Next, click on **Response** and specify the values you want to extract from the API response:
   ![response body attributes](/images/help/articles-img/help-articles/response-body-attributes.png)
   - **Attribute**: `room` — **Path**: `uuid`
   - **Attribute**: `greeting_message` — **Path**: `greeting_message`
5. Click **Go Back** to continue.

### 3.2: Save the Greeting Message

![save user response](/images/help/articles-img/help-articles/save-user-response.png)

1. Drag and drop a **Save user response** element into the flow.
2. Click the `{}` button and select the `greeting_message` attribute. (If you don't see it, go back to the **API** element and ensure the attribute is added correctly).
3. For **Save answer as**, create a new attribute called `body`.
4. In the **Fallback** section, select **Go to dialog** and choose **Default**.

### 3.3: Transition to the QA Dialog

![share attributes](/images/help/articles-img/help-articles/share-attributes-short.png)

1. Drag and drop the **To Dialog** element.
2. Select **QA** as the target dialog.
3. In the **Share attribute value(s) to next intent** section, click **ADD ATTRIBUTE** and select the following:
   - `room`
   - `body`
   - `greeting_message`

## Step 4: Configure the QA Dialog

![go to QA](/images/help/articles-img/help-articles/go-to-qa.png)

Now let’s configure the **QA** dialog, where AskHandle AI will provide responses.

Go to the **QA** dialog and add the following elements:
   - **API**
   - **Save user response**
   - **To Dialog**

### 4.1: Set Up API Request

1. Drag the **API** element into the QA dialog.
2. In the **API** element settings:
   - Select **POST** for the method.
   - ![message api](/images/help/articles-img/help-articles/message-api.png) 
     Paste the following URL: `https://dashboard.askhandle.com/api/v1/messages/`.
   
3. In the **Body** section, use the following JSON format:
   ```json
   {
     "body": "{{body}}",
     "nickname": "Guest",
     "room": { "uuid": "{{room}}" }
   }
   ```
   ![add header](/images/help/articles-img/help-articles/add-header-body.png) 

### 4.2: Set Up the Response Body

Click on **Response** and set up the response body attributes:
   - **Attribute**: `support_answer` — **Path**: `support_answer`
   ![support answers](/images/help/articles-img/help-articles/support-answers.png) 

### 4.3: Save the Response

1. Click on the **Save user response** element.
2. Click the `{}` button and select the `support_answer` attribute. ![save qa](/images/help/articles-img/help-articles/save-qa.png) 
3. For **Save answer as**, select `body`.
4. In the **Fallback** section, select **Go to Dialog** and choose **Default**.

### 4.4: Transition Back to Default Dialog

1. Go back to the **To Dialog** element.
2. Select **QA** as the target dialog. ![to qa](/images/help/articles-img/help-articles/to-qa.png) 
3. In the **Share attribute value(s) to next intent** section, click **ADD ATTRIBUTE** and select:
   - `room`
   - `body`
   - `support_answer`
   - `greeting_message`

Now your AskHandle AI is configured and ready to respond.

## Step 5: Test the AI

![simulator](/images/help/articles-img/help-articles/simulator.png) 

To test your AI setup:
1. Go to the **Test** section.
2. Click **START SIMULATION** to initiate a conversation with AskHandle AI.

## Step 6: Launch the AI

![set up sender](/images/help/articles-img/help-articles/set-up-sender.png) 

To launch your AI on a real WhatsApp number:
1. Go to **Settings** and select your WhatsApp number from the **Phone number** list.
2. If you haven't registered a WhatsApp number yet, navigate to **Channels and Numbers** -> **Numbers** and purchase a new number. Complete the WhatsApp registration process.

### Important Notice

![set up conversations](/images/help/articles-img/help-articles/set-up-conversations.png) 

If you intend to use AskHandle AI with a real WhatsApp number, ensure that **Conversations** is enabled in the **Default inbound configuration**.

If you don't see a Subscription ID, you should create a new one in the Developer Tools -> Subscriptions Management.


In this tutorial, we’ll guide you through the process of integrating AskHandle AI using the Infobip Chatbot builder. Infobip's Chatbot tool allows you to create chatbots for WhatsApp and SMS through an intuitive drag-and-drop interface. You can easily create rule-based chatbots or integrate generative AI capabilities directly into the builder.

How to Enable AskHandle AI on Infobip?


# How to Ensure Your AI Understands Webpage Content?

Enhancing your AI’s ability to process webpage content is a straightforward three-step process. First, we address potential issues that may arise when your AI attempts to scan webpages, such as restrictions that block content access. Next, we provide a practical solution by manually transferring webpage content into a format the AI can process more accurately. Finally, we explain how to test your AI’s comprehension after the upload and offer guidance on seeking additional support if needed.

## Step 1: Using Webpage as data source

![use web urls](/images/help/articles-img/help-articles/use-urls.png)

When you provide a webpage URL as data source, your AI tries to scan the page and learn from the content to generate responses. You can check the `Load URL recursively` option if you want the AI to learn all URLs from the link you provide. 

If your AI fails to deliver specific information or struggles to answer questions, it’s likely due to webpage restrictions on crawling. These restrictions prevent your AI from accessing the content it needs to provide accurate responses.

## Step 2: Implementing a Solution

To work around crawling restrictions, follow these steps to manually transfer webpage content into a document format:

1. **Copy the Webpage Content**: Go to the webpage, select all the relevant information your AI should understand, and copy it.
   
2. **Paste Into a Document**: Paste the copied content into a document format that your AI can easily process, such as a PDF, DOCX, or TXT file.

3. **Upload the Document**: In the Command Center, go to the file section and upload the document you created.

By uploading the content directly, you ensure your AI has access to the necessary information, improving its ability to understand and respond accurately.

## Step 3: Testing AI Recognition

Once the document is uploaded, test your AI to verify that it now provides relevant and specific answers based on the content from the uploaded document.

## Need Help?

If you encounter any issues or need further assistance, reach out to our support team at [support@askhandle.com](mailto:support@askhandle.com), and we’ll be happy to help.

Enhancing your AI’s ability to process webpage content is a straightforward three-step process. First, we address potential issues that may arise when your AI attempts to scan webpages, such as restrictions that block content access. Next, we provide a practical solution by manually transferring webpage content into a format the AI can process more accurately. Finally, we explain how to test your AI’s comprehension after the upload and offer guidance on seeking additional support if needed.

How to Ensure Your AI Understands Webpage Content?


# How to Customize AI Response?

Creating custom responses for your AI on the AskHandle dashboard is straightforward. The way your AI interacts with your customers largely depends on the content you provide and its structure. This process of training and setting up the AI is termed the "AI traning" phase. Essentially, it's about teaching the AI to provide specific information and communicate in your typical style.

Here’s a step-by-step guide:

## 1. Go to Command Center

Log in to your AskHandle user account and go to Command Center.

## 2. Select the Data Source Mode
Start by choosing your preferred Data Source mode. We provide four options: 
- **General Data Mode**
- **Your Own Data Mode**
- **Combined Data Mode**
- **Fine-tuned Model Mode**
- **Image Generator**

If you're uncertain about which mode to pick, refer to [our tutorial](choose-right-data-source-mode).

## 3. Add a Data Source

![Create AI answers](/images/help/articles-img/help-articles/add-data.png)

After selecting the Data Source Mode, click the "Add Data Source" button to begin fine-tuning your AI. We offer three methods to input your training material:

   - **Document:** Upload files in formats such as DOCX, PDF, XLSX, XSV, MD, TXT or JSON using the "Choose File" button or by dragging and dropping your files. We recommend adding one file at a time. If you'd like the AI to learn from multiple documents, create separate data sources for each.

   - **Plain Text:** Enter text directly by either pasting it into the plain text field or typing it manually. The AI will process this input and respond appropriately when customers ask related questions.
   
   - **Web page URL:** Direct the AI to a webpage for learning by specifying a URL. The AI will scan and process its content, but note that many websites—especially major ones like Instagram or StackOverflow—restrict AI access. This method works best for pages you manage or those with clear, unrestricted content. To include multiple pages from your website, you can enable the `Load URL recursively` option. The AI will then follow links from parent pages, depending on the site's structure, with a default depth of 3 levels. Alternatively, you can add different webpage URLs as separate data sources for more controlled AI responses.

## 4. Save the Data Source
After adding your materials, click the "Save Data Source" button.

## 5. Monitor the Processing Status
Post saving, you'll notice a status indicator. Initially, it will display `PROCESSING`, indicating your AI is assimilating the provided materials. This phase typically lasts between 10 and 20 seconds, though it may vary based on the volume of content. After 10-20 seconds have passed, simply refresh the page.

Once the AI completes its learning, the status will update to `PROCESSED`, indicating your AI ready to assist.


Creating custom responses for your AI on the AskHandle dashboard is straightforward. The way your AI interacts with your customers largely hinges on the content you provide and its structure. This process of training and setting up the AI is termed the Fine-Tuning phase. Essentially, it's about teaching the AI to communicate in your typical style.

How to Customize AI Response?

# How to Generate a BIMI-Compliant Logo with Adobe Illustrator?

Brand Indicators for Message Identification (BIMI) allows your logo to be displayed in email clients, helping to build trust and brand recognition. This guide explains how to create a BIMI-compliant logo using Adobe Illustrator and make the required changes to the SVG file.

## Step 1: Create or Open Your Logo in Adobe Illustrator

1. Open **Adobe Illustrator** and create a new document or load your existing logo.
2. Ensure the logo fits within a square canvas. Resize or adjust it if necessary.
3. Simplify the design by removing unnecessary effects or layers for better compatibility.

## Step 2: Export the Logo as an SVG

1. Save the finalized logo by selecting **File > Export > Export As**.
2. Choose **SVG** as the file format.
3. Use the following export settings:
   - **SVG Profiles**: Tiny 1.2
   - **Font**: Convert to Outlines
   - **Decimals**: 2
   - **Styling**: Inline Style
4. Save the file to generate the initial SVG version.

## Step 3: Adjust the SVG for BIMI Standards

After exporting the SVG file, it needs adjustments to meet BIMI standards. Open the file in a text editor and follow these steps.

### Example of Logo Code: Before and After

**Before Adjustments:**

```xml
<?xml version="1.0" encoding="utf-8"?>

<svg version="1.2" baseProfile="tiny" id="Слой_1"
     xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" x="0px" y="0px" viewBox="0 0 64 64"
     overflow="visible" xml:space="preserve">
<g>
    <path fill="#2B58F1" d="M62.7,22c-0.5-1.9-1.7-5.9-4.8-10c-6.5-8.7-16.4-11-18-11.3..."/>
    <path fill="#2B58F1" d="M34.1,16.8c-7.6-1.4-14.7,4.1-15.8,10.9c-1.1,6.9..."/>
</g>
</svg>
```

**After Adjustments:**

```xml
<?xml version="1.0" encoding="utf-8"?>

<svg version="1.2" baseProfile="tiny-ps" id="logo"
     xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink"
     viewBox="0 0 64 64" xml:space="preserve">
<title>Your Brand Logo</title>
<g>
    <path fill="#2B58F1" d="M62.7,22c-0.5-1.9-1.7-5.9-4.8-10c-6.5-8.7-16.4-11-18-11.3..."/>
    <path fill="#2B58F1" d="M34.1,16.8c-7.6-1.4-14.7,4.1-15.8,10.9c-1.1,6.9..."/>
</g>
</svg>
```

### Changes Made:
1. **Added a `<title>` Element**:  
   The `<title>` tag describes the logo, improving accessibility:
   ```xml
   <title>Your Brand Logo</title>
   ```

2. **Corrected `xml:space` Attribute**:  
   Ensured the attribute is set to `preserve` for proper XML compliance:
   ```xml
   xml:space="preserve"
   ```

3. **Removed Unnecessary Metadata**:  
   Removed redundant attributes like `overflow` and comments that are irrelevant for BIMI compliance.

4. **Simplified the ID Attribute**:  
   Updated the `id` attribute to a more descriptive value like `logo`.

5. **Validation**:  
   Validated the updated SVG using tools to ensure it meets BIMI standards.

## Step 4: Host the SVG on a Secure Server

Save the updated SVG file and host it on a secure HTTPS server. For example:
```
https://example.com/bimi/logo.svg
```

## Step 5: Update Your DNS Record

1. Log in to your DNS management system (e.g., AWS Route 53, Cloudflare).
2. Create a new TXT record:
   - **Name**: `default._bimi.example.com`
   - **Value**:  
     Without Verified Mark Certificate (VMC):  
     ```
     v=BIMI1; l=https://example.com/bimi/logo.svg;
     ```
     With VMC:  
     ```
     v=BIMI1; l=https://example.com/bimi/logo.svg; a=https://example.com/bimi/vmc.pem;
     ```

[How to Get a Verified Mark Certificate (VMC) for BIMI Compliance?](how-to-get-a-verified-mark-certificate-vmc-for-bimi-compliance)

## Step 6: Validate Your BIMI Record

Use tools like [BIMI Group’s Generator](https://bimigroup.org/bimi-generator/) to verify your setup. Confirm that:
- The DNS record is correct.
- The SVG is compliant with BIMI standards.

This step-by-step guide ensures your BIMI logo is created, adjusted, and implemented correctly.

How to Generate a BIMI-Compliant Logo with Adobe Illustrator?


# How to Get a Verified Mark Certificate (VMC) for BIMI Compliance?

Brand Indicators for Message Identification (BIMI) allows your logo to be displayed in email clients, helping to build trust and brand recognition. To enable BIMI, some email providers, such as Gmail, require a Verified Mark Certificate (VMC). This guide explains the steps to obtain a VMC and ensure your BIMI setup is compliant.

## What is a Verified Mark Certificate (VMC)?
A **VMC** is a digital certificate issued by a trusted Certificate Authority (CA). It verifies the authenticity of your logo and your organization, acting as proof that you own the rights to display the logo in emails. 

### Why is a VMC Important?
- **Required by Some Providers:** Gmail and other major email providers require a VMC for BIMI compliance.
- **Enhanced Trust:** It ensures only your organization can display the logo, preventing misuse or impersonation.
- **Trademark Verification:** A VMC requires your logo to be a registered trademark, providing additional credibility.

## Steps to Get a Verified Mark Certificate (VMC)

### 1. Register Your Logo as a Trademark
Your logo must be a registered trademark to qualify for a VMC. 
- Verify that your logo is registered with the appropriate trademark authority in your jurisdiction.
- Examples:
  - **United States**: U.S. Patent and Trademark Office (USPTO)
  - **European Union**: European Union Intellectual Property Office (EUIPO)

### 2. Choose a Certificate Authority (CA)
Select a trusted CA that offers VMC services. Two common providers are:
- [DigiCert](https://www.digicert.com/vmc/)
- [Entrust](https://www.entrust.com/digital-security/certificates/verified-mark-certificates)

### 3. Prepare the Required Documentation
CAs will require the following information to issue a VMC:
- **Trademark Registration**: Proof of your registered trademark.
- **Domain Verification**: Ownership of the domain used for your BIMI setup (via DNS records).
- **Organization Validation**: Documents such as business registration certificates to verify your legal entity.

### 4. Purchase the VMC
- Contact your chosen CA and follow their application process.
- Submit all requested documentation, including trademark and domain verification.

### 5. Host the VMC File
After approval, the CA will provide a PEM file for your VMC. Host this file securely on your domain, e.g.:
```
https://example.com/bimi/vmc.pem
```

## Update Your BIMI DNS Record to Include the VMC
Once you have the VMC, update your DNS record to point to both your BIMI-compliant logo and the VMC file.

### Example DNS Records
**Without VMC:**
```
v=BIMI1; l=https://example.com/bimi/logo.svg;
```
**With VMC:**
```
v=BIMI1; l=https://example.com/bimi/logo.svg; a=https://example.com/bimi/vmc.pem;
```

### Key Fields:
- `l`: The URL to your hosted BIMI-compliant logo.
- `a`: The URL to your hosted VMC file.

## Validate Your Setup
Use BIMI verification tools to confirm your DNS record and VMC compliance:
- [BIMI Group’s Generator](https://bimigroup.org/bimi-generator/)

## Frequently Asked Questions (FAQ)

### 1. Is a VMC required for all email providers?
No, not all email providers require a VMC, but providers like Gmail mandate it for BIMI compliance.

### 2. Can I use BIMI without a VMC?
Yes, some providers allow BIMI logos without a VMC, but you will be limited to providers that do not enforce this requirement.

### 3. How long does it take to get a VMC?
The timeline varies depending on the CA and the complexity of your application. Typically, it takes a few days to a couple of weeks.

Obtaining a Verified Mark Certificate (VMC) is a critical step for ensuring BIMI compliance with major email providers. Following these steps will help you successfully secure a VMC and display your logo, boosting brand recognition and trust in your emails.

How to Get a Verified Mark Certificate (VMC) for BIMI Compliance?


# Getting started with AskHandle

Welcome to AskHandle, where personalized AI support meets your needs! With our advanced AI platform and a user-friendly interface, we're here to streamline your experience and offer prompt, reliable assistance whenever you need it. Welcome aboard!

## 1. Log in to your AskHandle account:

After signing up for AskHandle, you have full access to your account dashboard. Go to https://dashboard.askhandle.com/accounts/login and log in with your username and password.

## 2. Know your dashboard:

In your user dashboard, you can access all available sections through the menu located on the left side. As you hover your mouse cursor over each section, the menu will expand, revealing the name of each section.

## 3. Configure the chat box visuals:

The Appearance section determines the visual presentation of your chat box on your website. AskHandle offers a variety of settings to customize the appearance of your chat box according to your preferences. To modify the default settings, navigate to the Appearance section and configure the appearance to suit your needs.

The key components within the Appearance section are options 1, 3, and 4. These settings form the fundamental framework of the chat box.

- **Option 1**: Greeting message
The greeting message is the initial message that AskHandle displays to users at the start of a new interaction. You can customize this message to include a welcoming phrase such as, "Hello, and welcome to ABC Company! How may we assist you today?"
- **Option 3**: Add a logo
The third option allows you to upload your logo. We recommend an image size of 170 x 170 pixels or a 1:1 aspect ratio. Your logo will appear in the pre-chat form and at the start of each interaction. Logos can be in PNG, JPEG, JPG, or BMP format.
- **Option 4**: Add a theme color
The theme color serves as the primary color of the chat box. You can effortlessly select a color that matches your style or utilize a HEX color code, such as #1f71ba. We advise against using RGBA color codes for this section, as they may occasionally cause color display issues on certain websites.
For further information on the Appearance section and configuring your chat box's appearance, please refer to our tutorial on customizing your chat box.

## 4. Import Information

Once you have completed the chat box configuration in Appearance, let’s go to the Command Center. The Command Center has two parts: Data Source and Interactive Options.

### Train the AI

Training your AI in the Command Center is a simple, yet powerful process that shapes the way your AI interacts with users and generates content. The key to effective training lies in the information you provide and how it's organized. The more structured and comprehensive the data, the more accurate and consistent your AI's responses will be. This phase, often referred to as "AI Training," is crucial in ensuring your AI understands not only the facts but also your preferred tone, style, and communication objectives.

To begin, you can create **custom prompts** tailored to guide the AI's responses. These prompts act as instructions, telling the AI what kind of answers it should provide and how it should phrase them. Whether you need the AI to adopt a formal tone or use casual language, a well-crafted prompt helps to align the output with your exact preferences. For example, you could instruct the AI to always answer in concise bullet points or to offer more detailed, conversational explanations depending on the context of the question.

In addition to prompts, the **Add Data Source** feature allows you to enhance the AI's knowledge base by incorporating your own documents, knowledge repositories, or proprietary data. Once added, the AI can seamlessly search through these documents to find relevant information, ensuring that it provides accurate answers based on the specific content you've uploaded. This is particularly useful for businesses that need their AI to respond with information that's relevant to their unique products, services, or internal processes. 

The combination of custom prompts and imported data sources means you have full control over the AI’s training. You’re not only teaching it the *what* (facts and information) but also the *how* (tone, style, and context), making it an invaluable tool for your specific needs. 

Once the training phase is complete, you can be confident that the AI will communicate in a way that aligns with your audience, delivering responses that are accurate, on-brand, and contextually relevant.

### Interactive Options

Interactive options are AI-powered quick-reply buttons that provide information with a simple click. Unlike static buttons, they offer dynamic responses. The text on each button acts as a query or prompt, which the AI uses to generate answers based on the fine-tuned data you've provided.

Locate the "Edit" button; it can be found next to the interactive options in the Command Center.

Click on the "Edit" button to open the interactive options' settings. Here, you can change the text on each button to better suit your needs. After editing the text, make sure to save your adjustments by clicking the "Save" button. This will ensure your changes are applied and visible to users.

If you’d like more information about training AI and configuring interactive options, please check the AskHandle tutorial on how to use the Command Center.

## 5. Test You AI Before Launch

You’re on the brink of launching your own AI agent. Before launching it publicly, it’s crucial to test its functionality thoroughly. The best way to accomplish this is by utilizing the Testing Page feature.

To access the Testing Page, navigate to its section and assign a name to the Page URL, such as "ABC Company." Save it and you will then get a unique link for your testing purposes. Additionally, you have the option to personalize the landing page by uploading your company logo and background image.

Once you've made these adjustments, simply click the “Save the Settings” button. Your Testing Page will be ready for immediate use and sharing. Click the provided link or scan the QR code to begin exploring and testing the capabilities of your AI.

## 6. Launch AI agent on Your Website

To install and launch your AI agent on your website, click the “Installation” button in the main menu. Copy & paste the code snippet to the index page of your website before the `</body>` tag, and the AskHandle messenger will be appeared on your website.

If you need any help with the installation, for example, if you have a difficulty with accessing the backend of your website where you would paste the install code, email us at support@askhandle.com or contact your customer success manager.


Welcome to AskHandle, where personalized AI support meets your needs! With our advanced AI platform and a user-friendly interface, we're here to streamline your experience and offer prompt, reliable assistance whenever you need it. Welcome aboard!

Getting started with AskHandle


# How to Get Your Token with AskHandle API?

This article guides you through the process of obtaining your AskHandle API token. The token lets you use our API to create chat rooms, send messages, and interact with your AI model.

## What is an API Token?

An API token is a unique key that authenticates your identity when using the AskHandle API. It is required to ensure secure access to our services.

## How to Get Your API Token

Follow the steps below to retrieve your API token:

### Step 1: Prepare Your Login Credentials

Make sure you have the following:

- Your AskHandle **username**
- Your AskHandle **password**

### Step 2: Generate Your Token Using `curl`

Open your terminal and enter the following command, replacing `your_username` and `your_password` with your AskHandle login details:

```sh
curl -X POST https://dashboard.askhandle.com/api/v1/auth/api-token \
 -H 'Content-Type: application/json' \
 -d '{"username": "your_username", "password": "your_password"}'
```

### Step 3: Retrieve Your Token

After executing the command, the API responds with your token in JSON format:

```json
{
  "token": "090de0abdfec0f17ca3c1bb27f91c3fb26171g05"
}
```

**Note:** This token is unique to your account and should be kept confidential. Do not share it publicly.

## How to Secure Your API Token

Your API token provides full access to your AskHandle resources. Keep it safe by:

- **Not sharing** it in public forums or unsecured channels.
- Storing it securely, ideally using encrypted storage or secret management tools.
- Rotating it regularly by generating a new token when necessary.

## What’s Next?

With your token ready, you can now:

- [Create chat rooms to start conversations](/how-to-build-a-chat-completion-experience-with-askhandle-api).
- Send messages to interact with AskHandle’s AI.
- Receive AI-generated responses to enrich user interactions.


This article guides you through the process of obtaining your AskHandle API token. The token lets you use our API to create chat rooms, send messages, and interact with your AI model.

How to Get Your Token with AskHandle API?


# How to Improve AI recognition for Lists and Characteristics?

When using AskHandle to provide detailed lists or characteristics about specific topics, it’s important to make sure the information is linked to the right subject. Here are simple steps to help the AI give better responses.

## Step 1: Organize Your Data

Make sure your data is clean and organized. If everything looks fine but issues still occur, move to the next step.

## Step 2: Clearly Mention the Subject
Always state the subject when presenting a list or set of details. If the subject is missing, the AI might connect the information to the wrong topic.

## Step 3: Use Clear References

Mention the subject when describing details so the AI knows what it is referring to.  
- **Correct**: *"Property A's features include X, Y, Z."*  
- **Incorrect**: *"Features include X, Y, Z."*

## Step 4: Be Specific with Attributes

If listing details for multiple items, make sure to clearly mention which item each attribute belongs to.  
- **Correct**: *"Unit sizes for Property A range from studio to one-bedroom."*  
- **Incorrect**: *"Unit sizes range from studio to one-bedroom."*

## Step 5: Avoid Vague Statements

Avoid general statements that don’t specify a subject. Always connect details directly to the topic you're describing.

Following these steps will help AskHandle understand your data better and provide more accurate responses.

When using AskHandle to provide detailed lists or characteristics about specific topics, it’s important to make sure the information is linked to the right subject. Here are simple steps to help the AI give better responses.

How to Improve AI recognition for Lists and Characteristics?


# How to Improve AI Recognition in Other Languages?

AskHandle AI supports over 90 languages and can communicate with your customers in any of them. By default, the AI can understand and process information from your training data regardless of its original language, generating responses in the language your customers prefer.

However, languages often have unique nuances, especially when it comes to localized terms, brand names, or cultural expressions. If AskHandle’s recognition or content generation does not meet your expectations in a specific language, you have two options to improve its performance:  

1. **Upload relevant documents as a data source**: With retrieval-augmented generation (RAG), the AI can search through your uploaded documents to generate more accurate and context-aware responses.  
   
2. **Create a fine-tuned AI model**: If you're working with a lesser-known or small language, you can upload data in that language to fine-tune the AI model. This allows the AI to better understand subtle linguistic nuances and domain-specific terms, improving response accuracy and enhancing customer interactions.

By implementing these options, you can optimize AskHandle’s ability to provide high-quality, localized answers tailored to your customers’ needs.



AskHandle AI supports over 90 languages and can communicate with your customers in any of them. By default, the AI can understand and process information from your training data regardless of its original language, generating responses in the language your customers prefer.

How to Improve AI Recognition in Other Languages?


# How to Insert Images in AI Responses?

Incorporating images in AskHandle AI responses enhances user engagement by providing visual context alongside text. Follow this guide to automate image inclusion in your AI responses based on specific keywords.

## Step 1: Prepare the Image Sources

Start by creating a **CSV file** that contains keywords and corresponding image URLs. For example, if your AI handles questions about New York tourist attractions, you can link images to keywords like "Manhattan," "Central Park," and "Statue of Liberty."

Your CSV file should have two columns:
- **Keyword**: The word or phrase that triggers image insertion.
- **Picture URL**: The direct Unsplash or image link.

Here’s an example CSV:

| Keyword           | Picture URL                                                |
|-------------------|------------------------------------------------------------|
| Manhattan         | https://images.unsplash.com/photo-1485738422979-f5c462d49f74 |
| Times Square      | https://images.unsplash.com/photo-1533577116850-9cc66cad8a9b |
| Central Park      | https://images.unsplash.com/photo-1528747045269-390fe33c19e1 |
| Brooklyn Bridge   | https://images.unsplash.com/photo-1559582792-cb2137b2ad63   |
| Statue of Liberty | https://images.unsplash.com/photo-1559170054-0762eaff38c2   |


## Step 2: Upload the CSV File

Once your CSV is ready, upload it to the AskHandle system. The AI will use this file to match keywords in user queries with image URLs.


## Step 3: Guide the AI for Image Handling

Create a text guide in `custom prompt` to instruct the AI on how and when to insert images. Define that the AI should check for keywords and display the corresponding image before the text response using Markdown format.

**Example Guide:**  
If the AI detects keywords such as "Manhattan," "Times Square," or "Central Park," it should:
1. Retrieve the related image from the CSV file.
2. Insert the image before the text response using the Markdown format.  
   Example:  
   ```markdown
   ![Manhattan](https://images.unsplash.com/photo-1485738422979-f5c462d49f74)
   ```

## Step 4: Integrate Files into AI

With the uploaded CSV file and your custom prompt, the AskHandle AI will be trained to:  
1. **Search for Keywords**: Scan user queries for keywords in the CSV file.  
2. **Insert Pictures**: Retrieve and display the associated image using Markdown syntax.  
3. **Format the Response**: Ensure the image appears above the text response, sized with the `w=230` parameter for consistency.


## Step 5: Test AI Responses

Test the AI to verify that images are inserted correctly. Below are sample responses:

### Example 1: Query about Manhattan

```
![Manhattan](https://images.unsplash.com/photo-1485738422979-f5c462d49f74?w=230)

Manhattan is known for its iconic skyline, towering skyscrapers, and landmarks like Times Square and Central Park.
```

### Example 2: Query about Brooklyn Bridge

```
![Brooklyn Bridge](https://images.unsplash.com/photo-1559582792-cb2137b2ad63?w=230)

The Brooklyn Bridge connects Manhattan and Brooklyn, offering stunning views of the New York City skyline.
```

### Example 3: Query about Statue of Liberty

```
![Statue of Liberty](https://images.unsplash.com/photo-1559170054-0762eaff38c2?w=230)

The Statue of Liberty symbolizes freedom and democracy, standing tall on Liberty Island in New York Harbor.
```

By following this guide, you can enhance AI responses with visuals, providing users with a more engaging and informative experience.

Incorporating images in AskHandle AI responses enhances user engagement by providing visual context alongside text. Follow this guide to automate image inclusion in your AI responses based on specific keywords.

How to Insert Images in AI Responses?


# How to Insert URLs in AI Responses?

Including URLs in AI responses helps users access detailed information and additional resources. This guide explains how to configure AskHandle to show URLs in responses.

### Step 1: Upload Documents with URLs

Start by uploading documents that contain URLs linked to specific answers. These documents form the AI’s searchable knowledge base.  

For example, if a user asks, *“Do you have any tourist routes to New York?”*, the document might contain the following text:  
> **Our New York tour details:** Visit [our website](https://www.xyz.com/new-york-tour) for more information on routes and attractions.

AskHandle AI will find and display this link when generating answers based on the document's content.


### Step 2: Write a Custom Prompt for the AI

To ensure that URLs are consistently included in responses, provide a clear instruction in the **Custom Prompt** section.  

A sample prompt could be:  
> **“You should include a link in the answers.”**  

This prompt helps guide the AI to prioritize relevant links when responding to user questions.

### Step 3: Embed URLs in Text Context

Embedding URLs naturally within sentences improves AI recognition and response quality. For example, your document might include a sentence like:  
> **Our New York tour includes five tourist attractions. For more details, visit:** [https://www.xyz.com/new-york-tour](https://www.xyz.com/new-york-tour).

This approach helps ensure that the AI provides both context and useful links in its answers.



Including URLs in AI responses helps users access detailed information and additional resources. This guide explains how to configure AskHandle to show URLs in responses.

How to Insert URLs in AI Responses?

# How to install AskHandle chat widget on a single-page website?

If you're building your website using ReactJS or another single-page application (SPA) framework, installing the chat widget requires a slightly different approach than a traditional website. Because SPAs load content dynamically without full page reloads, embedding scripts like our widget can fail if not configured correctly.

This guide walks you through the recommended installation methods, highlights common mistakes (especially when using tools like Google Tag Manager), and provides working code examples to ensure your widget displays as expected on every route.


## ✅ Method 1: Add to `public/index.html` (Most Reliable)

This is the best option **if your project has access to `index.html`** (e.g., Create React App).

**Steps:**

1. Open your React project.
2. Edit `public/index.html`.
3. Paste this just before `</body>`:

```html

<script>
  window.webchatConfig = { token: "N3623213-309-356-0797850M" };
</script>
<script src="https://handle-chat-widget.s3.amazonaws.com/assets/js/webchat-widget.min.js"></script>
```

✅ This ensures the config is set before the script loads, and only loads once.


## ⚠️ Method 2: Google Tag Manager — Use With Caution

### 🚫 Common Mistake

* You add the script in a **Custom HTML tag** with a **Page View** trigger.
* This causes the script to load **before `window.webchatConfig` is available**.
* Result: **Widget silently fails** — no error, no chat.

### ✅ Correct Setup

1. In GTM, create a **Custom HTML Tag**.
2. Paste:

```html
<script>
  window.webchatConfig = { token: "N3182213-309-356-0797850M" };
</script>
<script src="https://handle-chat-widget.s3.amazonaws.com/assets/js/webchat-widget.min.js"></script>
```

3. Set the trigger to:
   **✔️ “DOM Ready”** (not "Page View")

✅ This ensures the script loads at the right time, with the config available.


## 🛠️ Method 3: Inject Widget in React (If No `index.html`)

If your site **doesn’t use `index.html`** — like in **Next.js**, **Remix**, or headless CMS setups — use this React-based approach:

```jsx
import { useEffect } from 'react';

function useChatWidget() {
  useEffect(() => {
    // Prevent duplicate loads
    if (document.getElementById('chat-widget-script')) return;

    // Set config
    window.webchatConfig = { token: 'N3182213-309-356-0797850M' };

    // Inject script
    const script = document.createElement('script');
    script.id = 'chat-widget-script';
    script.src = 'https://handle-chat-widget.s3.amazonaws.com/assets/js/webchat-widget.min.js';
    script.async = true;
    document.body.appendChild(script);
  }, []);
}

export default function App() {
  useChatWidget();

  return (
    <div>
      <h1>My App</h1>
      {/* Rest of your app */}
    </div>
  );
}
```

> 🧠 Place this in your root component (like `_app.js` or `Layout.jsx`) to ensure it loads once.

## 🔁 Route Changes in SPAs

SPAs like React don't reload the page when navigating — so your widget only needs to load **once**, as long as you don’t remove it from the DOM. Avoid reloading the widget script on every route change.

## ✅ Summary

| Method           | Use Case                              | Safe for SPA? | Notes                          |
| ---------------- | ------------------------------------- | ------------- | ------------------------------ |
| `index.html`     | Create React App, static SPA          | ✅ Yes         | Most stable method             |
| GTM w/ Page View | Any SPA                               | ❌ No          | Config not ready; widget fails |
| GTM w/ DOM Ready | Any SPA                               | ✅ Yes         | Must define config first       |
| React hook       | Next.js, headless CMS, no HTML access | ✅ Yes         | For advanced setups            |

If you're building your website using ReactJS or another single-page application (SPA) framework, installing the chat widget requires a slightly different approach than a traditional website. Because SPAs load content dynamically without full page reloads, embedding scripts like our widget can fail if not configured correctly.

How to install AskHandle chat widget on a single-page website?


# How to Install AskHandle on a Next.js Website?

Next.js is a popular React-based framework used to build dynamic web applications. Unlike standard websites, Next.js doesn't have a simple `index.html` file, so you'll need to adjust the AskHandle widget installation slightly. This article provides clear steps and example code for adding the AskHandle chat widget to your Next.js website.

## Where Should You Add the Widget Code in Next.js?

In Next.js, the best place to insert scripts like the AskHandle widget is inside the `_app.js` file. This ensures the widget loads consistently across every page of your application.

## Next.js `Script` Component

Next.js includes a built-in component called `Script` to manage external JavaScript. It allows efficient loading and offers several loading strategies:

- **`afterInteractive` (recommended)**: Loads scripts immediately after the page renders.
- **`beforeInteractive`**: Loads scripts before the page renders.
- **`lazyOnload`**: Loads scripts after the entire page loads.

We'll use the default and recommended strategy, `afterInteractive`, for the AskHandle widget.

## Step-by-Step Installation Guide

Follow these steps to integrate the AskHandle chat widget into your Next.js website.

### **Step 1: Locate or Create `_app.js`**

Find your `_app.js` file inside the `pages` folder:

```
your-nextjs-app/
├── pages/
│   ├── _app.js   <-- Modify or create this file
│   ├── index.js
├── components/
├── public/
└── styles/
```

If `_app.js` doesn’t exist, create a new one in the `pages` directory.

### **Step 2: Open `_app.js` in Your Code Editor**

Open the `_app.js` file using your preferred editor (e.g., Visual Studio Code).

### **Step 3: Import `Script` Component from Next.js**

At the top of `_app.js`, add the following import statement:

```jsx
import Script from 'next/script';
```

### **Step 4: Add the AskHandle Widget Installation Code**

Copy and paste the following example code inside your `_app.js`. Replace the token in `"token": "YOUR_TOKEN_HERE"` with the actual token from your AskHandle dashboard (available under the **Installation** section):

```jsx
export default function App({ Component, pageProps }) {
  return (
    <>
      <Script id="askhandle-config" strategy="afterInteractive">
        {`
          window.webchatConfig = {
            "token": "YOUR_TOKEN_HERE"
          };
        `}
      </Script>

      <Script
        src="https://handle-chat-widget.s3.amazonaws.com/assets/js/webchat-widget.min.js"
        strategy="afterInteractive"
      />

      <Component {...pageProps} />
    </>
  );
}
```

Make sure to replace `"YOUR_TOKEN_HERE"` with your unique token. Here's how your final code might look with an example token:

**Example:**
```jsx
export default function App({ Component, pageProps }) {
  return (
    <>
      <Script id="askhandle-config" strategy="afterInteractive">
        {`
          window.webchatConfig = {
            "token": "M7055559-408-649-0406487A"
          };
        `}
      </Script>

      <Script
        src="https://handle-chat-widget.s3.amazonaws.com/assets/js/webchat-widget.min.js"
        strategy="afterInteractive"
      />

      <Component {...pageProps} />
    </>
  );
}
```

### **Step 5: Save and Test Your Changes**

After saving your changes, restart your Next.js development server:

```sh
npm run dev
```

Open your website (typically at `http://localhost:3000`) and verify that the AskHandle chat widget appears correctly at the bottom right corner of all pages.

## Troubleshooting Tips

If the widget doesn’t appear:

- Verify your API token is correct.
- Check browser console logs for errors.
- Ensure your Next.js server has been restarted.

You've successfully integrated the AskHandle chat widget into your Next.js website.

Next.js is a powerful React-based framework commonly used to build dynamic web applications. Unlike traditional HTML-based sites, Next.js does not have a standard index.html file where scripts can simply be pasted. Because of this, installing the AskHandle chat widget requires minor adjustments to fit the Next.js structure.

How to Install AskHandle on a Next.js Website?


# How to Install AskHandle on a WordPress Website?

Installing AskHandle on your WordPress website is simple and straightforward. The fastest method is using a WordPress plugin, which allows you to easily add the necessary JavaScript snippet to your website. Follow these steps to install AskHandle:

## Step 1: Install the Plugin

![install plugin](/images/help/articles-img/help-articles/wordpress1.png)

1. Go to the `Plugins` section of your WordPress dashboard.
2. Click on the `Add New` button.
3. In the search bar, type in **"Head and Footer Scripts Inserter"** (or search for any plugin that helps you insert scripts into the header or footer).
4. You'll see several plugins listed. Choose **"Head and Footer Scripts Inserter"** (or your preferred plugin) from the search results.
5. Click **Install Now**, then click **Activate** once the installation is complete.

The **Head and Footer Scripts Inserter** plugin allows you to easily insert custom scripts, like the one required to integrate AskHandle, to every page of your WordPress site.

## Step 2: Configure the Plugin to Install AskHandle

![install chat box](/images/help/articles-img/help-articles/wordpress2.png)

1. After activating the plugin, go to the `Plugins` menu again and locate the **Head and Footer Scripts Inserter** plugin.
2. Click on **Settings** under the plugin name.
3. On the settings page, you'll see a field labeled **"Scripts in Footer"**. This is where you'll paste the AskHandle installation snippet.
4. Copy the JavaScript code provided to you by AskHandle and paste it into the **"Scripts in Footer"** field, just before the closing `</body>` tag.
5. Click the **Save Changes** button to apply the script to your website.

Now, the script will be added to the footer of every page on your site, allowing the AskHandle chatbox to appear.

## Step 3: Verify the Installation

1. Visit any page on your website to see if the AskHandle chatbox appears in the bottom-right corner.
2. If the chatbox doesn’t show up immediately, try clearing your browser cache or refreshing the page.

And that's it! You’ve successfully added AskHandle to your WordPress site. The chatbot should now be ready to assist your visitors.


Installing AskHandle on your WordPress website is simple and straightforward. The fastest method is using a WordPress plugin, which allows you to easily add the necessary JavaScript snippet to your website. Follow these steps to install AskHandle:

How to install AskHandle on a Wordpress website?


# How to Install AskHandle on Your Website?

Installing the AskHandle chat widget on your website is quick and easy. Follow these simple steps to get started:

## Step 1: Add Your Website Domain(s)

First, specify the domain(s) where you'll be using AskHandle. This step links the AskHandle widget securely to your website.

- Log in to your AskHandle account.
- Navigate to the **Installation** section of your dashboard.
- Enter the domain(s) for your website.

**Example:** If your website is accessible at `https://www.example.com`, we recommend entering both:

```
https://www.example.com, https://example.com
```

**Note:**
- Separate multiple domains using commas.
- If you're using AskHandle on other subdomains, include them as well. For example:

```
https://app.example.com
```

- After entering your domains, click **Save the domain**.

## Step 2: Copy and Paste the Installation Code

Next, copy the AskHandle installation code provided in your dashboard and paste it into your website.

Place the following code snippet right before the closing `</body>` tag on every page where you want the AskHandle widget to appear:

```html
<script>
  window.webchatConfig = {"token": "M5835529-408-648-0403487G"}
</script>
<script type="text/javascript" src="https://handle-chat-widget.s3.amazonaws.com/assets/js/webchat-widget.min.js"></script>
```

**Important:** Replace `"M5835529-408-648-0403487G"` with the actual token from your AskHandle dashboard.

## Step 3: Verify Installation

Once you've added the code snippet to your website, save your changes and refresh your site. You should now see the AskHandle chat widget active and ready to engage your visitors.

If the widget doesn't appear immediately, double-check the following:

- Ensure the token in the installation code matches the one provided in your dashboard.
- Confirm that your domain(s) have been correctly saved.

That's it! You've successfully installed AskHandle on your website.



Installing the AskHandle chat widget on your website is quick and easy. Follow these simple steps to get started.

How to Install AskHandle on Your Website?


# How to measure AI Performance?

At AskHandle, you can evaluate AI performance by ensuring responses are accurate and relevant to user questions. Each AI-generated response is accompanied by two key scores, shown in brackets (e.g.,`(0.8, 0.9)`), that reflect the quality of the AI’s performance. These scores, ranging from 0 to 1, provide a quick snapshot of how effectively the AI meets user needs.

- **The first score** is the **Faithfulness Score**: This checks if the AI’s response is factually consistent with the data you’ve provided.
- **The second score** is the **Relevance Score**: This indicates how well the response addresses the specific question asked.

If either score is zero, meaning the AI couldn’t use your data to respond, you’ll see “General Data” displayed instead of the scores, letting you know the AI used general knowledge to answer the question.

## Understanding the Faithfulness Score

The [Faithfulness Score](/what-is-faithfulness-score-in-ai-responses) (the first number) measures how closely the AI’s response sticks to the factual information in your data. This score ranges from 0 to 1:

- **Score of 1**: The AI response is fully aligned with the facts in your data.
- **Score of 0**: The response does not use any facts from your data.
- **Scores between 0 and 1**: The response is partially accurate, where some claims match your data, but there might be slight differences or additions.

For instance, if a customer asks about your store’s opening hours and the AI’s response matches the exact hours in your data, you’ll see a high Faithfulness Score. This score ensures the AI’s response is consistent with the information you’ve provided.

## Understanding the Relevance Score

The [Relevance Score](/what-is-ai-answer-relevance-score) (the second number) measures how well the AI’s response answers the specific question being asked. This score also ranges from 0 to 1:

- **Score of 1**: The AI’s response directly answers the user’s question using the data.
- **Score of 0**: The response does not address the question with your data.
- **Scores between 0 and 1**: Partial relevance, where the response is somewhat related to the question but doesn’t fully address it.

For example, if a user inquires about product availability, a high Relevance Score means the AI’s response is directly related to availability based on your data. This score confirms that the AI is responding accurately to the user's specific question.

## Why Faithfulness and Relevance Scores Matter in Measuring AI Performance

Together, these two scores provide insight into the AI’s performance:

- **Faithfulness Score**: Shows if the AI uses your data accurately.
- **Relevance Score**: Confirms if the AI is addressing the user’s question specifically.

High scores in both Faithfulness and Relevance indicate a high-quality response. Low scores in either area might suggest gaps in your data or areas where the AI might need additional context or refinement.

## Improving AI Performance with These Scores

Monitoring Faithfulness and Relevance Scores allows for ongoing improvement of AI performance. Low Faithfulness Scores may indicate that the AI is making assumptions or adding unnecessary details, suggesting a need for clearer or more precise data. Low Relevance Scores imply the AI could benefit from more specific information to address particular questions, guiding you to refine or expand your data sources.

Faithfulness and Relevance Scores offer a straightforward approach to measuring and enhancing your AI’s performance. This oversight ensures that your AI chatbot not only responds quickly but also provides factually accurate and relevant answers, creating a more satisfying and trustworthy customer experience in every interaction.

At AskHandle, we assess AI performance by ensuring that responses are both accurate and relevant to the user's question. Each AI-generated response in AskHandle includes two key scores, displayed in brackets like (0.8, 0.9), that measure the quality of the AI’s performance. These scores range from 0 to 1 and give a quick indication of how well the AI meets user needs.

How to measure AI Performance?


# How to Provide General Responses to Customer Questions?

Customers may sometimes ask questions that are not directly related to your services or outside the scope of the data you’ve uploaded. You can guide the AI to respond appropriately by adding detailed information to your data source and using custom prompts for further control.

## Step 1: Add General Information to Your Data Source

Include a general text response with key information, such as contact details, in your uploaded data. This allows the AI to reference the information when answering broad or unrelated questions.  

For example, you could add a statement like:  
*"For more information or to contact us directly, please call 123-456-7890 or email us at info@company.com."*  

When the AI finds this in the data source, it can use it to provide helpful responses to general questions.

## Step 2: Use a Custom Prompt for More Control

Create a custom prompt to instruct the AI on how to handle questions that are unrelated to your services. For example, you can add a prompt such as:  
*"If a question is not related to our services, provide a response that includes our contact details."*  

This allows you to tailor how the AI handles queries not covered by your uploaded data.

## Step 3: Test and Adjust the AI’s Responses

Run tests to check how the AI handles a variety of questions. If responses are inconsistent, update your data source or custom prompts to improve the AI’s performance.  

Including relevant data and instructions helps ensure that your AI provides clear, helpful responses, even when customers ask questions outside its specialized knowledge. This creates a more reliable and positive user experience.

Customers may sometimes ask questions that are not directly related to your services or outside the scope of the data you’ve uploaded. You can guide the AI to respond appropriately by adding detailed information to your data source and using custom prompts for further control.

How to Provide General Responses to Customer Questions?


# How to Set Up a Self-Introduction for Your AI

Setting up a proper self-introduction helps your AI assistant create a welcoming and professional impression. You can achieve this by providing a clear instruction in the **Custom Prompt** section.

## Step 1: Write a Self-Introduction Instruction

In the **Custom Prompt**, define how the AI should introduce itself to users. For example:

**Prompt Example:**  
*"You are AskHandle AI assistant. Your name is Mary, and you work for AskHandle. When users ask who you are, introduce yourself politely."*  

This instruction ensures the AI introduces itself consistently. Remember to save your changes after writing the prompt.

Depending on your business needs, you can modify the introduction to fit your brand. Here are a few examples:

1. **Basic Introduction:**  
*"You are the customer support assistant for XYZ Company. Your name is Alex, and your role is to help customers with their inquiries."*

2. **Friendly Tone:**  
*"You are the AskHandle AI assistant named Lucy. You are here to help answer any questions. Feel free to mention you're happy to assist!"*

3. **Formal Introduction:**  
*"You are a virtual assistant representing AskHandle. Your name is John. Kindly greet users and offer professional assistance."*

4. **Role-Specific:**  
*"You are the technical support assistant for Tech Solutions Inc. Your name is Emma. Introduce yourself and explain that you assist users with tech-related issues."*

## Step 2: Test the AI’s Introduction

After setting up the custom prompt, ask the AI questions like *“Who are you?”* or *“Can you introduce yourself?”* to verify that it provides the correct response. Update the prompt if needed to improve clarity.

Setting a proper introduction helps users feel more comfortable and establishes your AI's role in assisting them.

Setting up a proper self-introduction helps your AI assistant create a welcoming and professional impression. You can achieve this by providing a clear instruction in the Custom Prompt section.

How to Set Up a Correct Self-Introduction for Your AI?


# How to Set Up AskHandle AI for Scheduling Appointments?

Setting up AskHandle AI to manage scheduling for your users is simple. Follow these steps to allow the AI to handle appointments, ask necessary questions, and confirm visits.

### Step 1: Prepare a Word Document with Scheduling Details

To guide the AI in handling appointment requests, you can upload a Word document template to the AskHandle Command Center. Below is an example template that you can copy, customize, and upload:

**Example AI Training Template for Scheduling Appointments:**

---

If a guest user wants to schedule a visit or book a tour, you should help get an available date and time. You can ask the guest user what’s the ideal date and time the guest user wants to come and visit. 

Our business hour is 9 am to 5 pm, Monday to Friday. And we don’t work on Federal holidays. We use Eastern Time as our time zone.

And our showing manager’s available time for the coming days are:

- Monday: 9 am to 11 am, 1 pm to 4 pm
- Tuesday: 10 am to 11:30 am, 12 pm to 5 pm
- Wednesday: 9 am to 12 pm, 3 pm to 4 pm
- Thursday: 10:30 am to 4 pm
- Friday: 1 pm to 3 pm

If the guest user asks for a time that no show manager is available, you can ask the guest user to tell a different date and time.

If the guest user has a time that is in our available date and time range, you should ask about guest user email.

If the guest user gives an email, you should repeat the email to ask guest user to confirm if the email is correct.

After you get available date, time and confirmed email, you can tell the date, time, email you received and you should ask the guest user to type “confirm the visit” to confirm the visit.

After you receive the confirm the visit message from guest user, you can tell the guest user “Your visit is confirmed.” And you should mention the confirmed date and time in the confirmation message.

---

### Step 2: Upload the Document to AskHandle Command Center

Upload the document to the [Command Center](https://dashboard.askhandle.com/stories/command_center/):

- Go to the AskHandle Command Center.
- Upload the Word document with the instructions for handling appointments.
- This document will tell the AI how to interact with the user when scheduling visits.

### Step 3: Set Available Time Slots in AskHandle

Set available time slots:

- Ensure you set the available time slots in the AskHandle system.
- Input your showing manager’s availability to allow the AI to suggest open slots.
- This ensures the AI only offers times when your staff is available.

### Step 4: Test the AI Scheduling Process

After uploading the document and setting time slots, test the AI to make sure it follows the correct process.

The AI should:
- Ask users for their preferred date and time.
- Compare the user’s choice with your available time slots.
- Request and confirm the user’s email.
- Complete the appointment confirmation process.

By following these steps and using the provided template, your AskHandle AI will efficiently manage user scheduling, making it easier to book visits or tours within your available hours.

Setting up AskHandle AI to manage scheduling for your users is simple. Follow these steps to allow the AI to handle appointments, ask necessary questions, and confirm visits.

How to Set Up AskHandle AI for Scheduling Appointments?


# How to Set Up the Greeting Message?

When users start a conversation with AskHandle, they are greeted with a welcoming message. You can easily customize this greeting to better match your brand's tone.

## Customizing the Greeting

To set up or edit your chatbot's greeting message:

1. Go to the **[Appearance](https://dashboard.askhandle.com/customization/)** section in the dashboard.
2. In the **Appearance** settings, the first option allows you to edit the greeting message. Simply type your desired text in the provided box and click **Save** to apply the changes.

![Greeting menu](/images/help/articles-img/help-articles/greeting-menu.png)

## Adding a Personal Touch

For a more personalized greeting, you can use `{username}`. For example, you could write:

`Hi {username}, how can I assist you today?`

If the user’s name is available, it will automatically be inserted, creating a more tailored experience.



When users start a conversation with AskHandle, they are greeted with a welcoming message. You can easily customize this greeting to better match your brand's tone.

How to Set Up the Greeting Message?


# How to Set Up Interactive Options?

Interactive options are quick-reply buttons that let users click to send short messages. You can use these buttons to highlight frequently asked topics or guide users through common actions.

## Editing Interactive Options

1. In the **Command Center**, locate the **"Edit"** button next to the interactive option you want to modify.
   
2. Click the **"Edit"** button to open the settings for that button. Here, you can update the text to better match your needs.

3. After making your changes, click **"Save"** to apply and update the button.

## Removing Interactive Options

1. Click the **"Edit"** button next to the interactive option you want to remove.

2. In the settings, uncheck the **"Show Button"** box to hide the button.

3. Click **"Save"** to finalize the removal. The button will no longer appear to users.

Interactive options are designed to improve user engagement. Keep in mind that changing or removing them can impact how users interact with your AI.


Interactive options are quick-reply buttons that let users click to send short messages. You can use these buttons to highlight frequently asked topics or guide users through common actions.

How to Set Up Interactive Options?


# How to Subscribe to an RSS Feed?

RSS feeds offer a streamlined solution to receive updates without having to visit each site individually. RSS, which stands for Really Simple Syndication, uses a family of standard web feed formats to publish frequently updated information such as blog entries, news headlines, audio, and video. In this article, we'll explore how you can subscribe to an RSS feed, specifically focusing on the RSS feed provided by [AskHandle](https://www.askhandle.com/blog-rss.xml), ensuring you never miss an update from their insightful blog.

## Understanding RSS Feeds

RSS feeds allow users to easily stay informed by retrieving the latest content from the sites they are interested in. You can view this information in one place—your RSS feed reader. This saves time by not needing to visit each site individually and ensures you do not miss any updates.

### Benefits of Using RSS

- **Efficiency**: RSS saves time and effort by eliminating the need to manually check each website for updates.
- **Simplicity**: RSS feeds are simple to use and easy to set up.
- **Customization**: Users can tailor which content they receive updates for, ensuring they only get the most relevant information.
- **Aggregation**: All news comes to one spot, making it easier to keep track of diverse content sources.

## How to Access RSS Feeds

To access RSS feeds, you need an RSS feed reader, also known as an aggregator. These readers are available as dedicated websites, desktop applications, or mobile apps. Some popular RSS readers include Feedly, Inoreader, and The Old Reader. Here's a step-by-step guide on how to subscribe to an RSS feed using one of these tools.

### Step 1: Choose an RSS Reader

First, select an RSS reader. Each reader comes with its own set of features, so choose one that best fits your needs. For beginners, web-based readers like Feedly or Inoreader are recommended due to their user-friendly interfaces and robust feature sets.

### Step 2: Locate the RSS URL

For the purpose of this tutorial, we will subscribe to the AskHandle blog. The RSS URL you will need is: [https://www.askhandle.com/blog-rss.xml](https://www.askhandle.com/blog-rss.xml).

### Step 3: Add the RSS Feed to Your Reader

The process for adding a feed varies slightly depending on the reader, but generally, it involves these steps:
1. Log into your RSS reader.
2. Look for an option that says "Add Feed" or "Subscribe to Feed."
3. Enter the RSS feed URL: `https://www.askhandle.com/blog-rss.xml`.
4. Confirm the addition and adjust settings such as update frequency if applicable.

### Step 4: Organize Your Feeds

Most RSS readers allow you to organize feeds into categories or folders. If you follow multiple feeds, organizing them can help keep your information manageable and sorted by topic or priority.

## Tips for Managing Your RSS Subscriptions

- **Regularly Review Your Feeds**: Periodically review your subscriptions to remove those you no longer find useful or interesting.
- **Categorize Your Feeds**: Use categories or tags in your RSS reader to organize feeds according to topics. This can be particularly helpful if you subscribe to many different sources.
- **Set Up Notifications**: Some RSS readers allow you to set up notifications for feeds with particularly important updates. Use this feature sparingly to avoid notification overload.

## Advantages of RSS for Content Consumers

RSS feeds provide a level of customization and control over the content you receive that is unmatched by traditional methods of online media consumption. They can be particularly beneficial for:
- **Students and researchers** who need regular updates on specific topics.
- **Professionals** who need to keep abreast of industry news.
- **Anyone** who wants to stay informed about their interests without the hassle of bouncing from site to site.

Subscribing to an RSS feed simplifies how you access updates, ensuring you're always informed with minimal effort. RSS feeds represent a powerful tool in managing the information overload in today's digital world. By following the steps outlined above, you can efficiently start incorporating RSS into your daily routine, saving time and staying informed with ease.

RSS feeds offer a streamlined solution to receive updates without having to visit each site individually. RSS, which stands for Really Simple Syndication, uses a family of standard web feed formats to publish frequently updated information such as blog entries, news headlines, audio, and video. In this article, we'll explore how you can subscribe to an RSS feed, specifically focusing on the RSS feed provided by [AskHandle](https://www.askhandle.com/blog-rss.xml), ensuring you never miss an update from their insightful blog.

How to Subscribe to an RSS Feed?


# Teaching AskHandle AI to Use First Person

AskHandle AI reflects the data you've provided during its training. If your training data mostly uses third-person language, the AI might respond in third person, even when using the first person would be more appropriate.

To ensure the AI uses first-person responses, you can write custom prompts that guide the AI to answer in the first person. This makes interactions feel more personal and engaging.

## How to Teach the AI to Use First Person

Simply use the following instructions in your prompts to guide the AI’s response:

### 1. Basic First-Person Prompt  

**Prompt:** "Use first person like 'I' to answer all questions in the prompt."

**Result:** The AI will now use "I" to respond to customers, creating a more personal interaction.

### 2. Example of Personalizing Support

**Prompt:** "When responding to support inquiries, always use 'I' instead of 'we' or 'the company.' For example, say 'I will help you with that,' instead of 'We will help you with that.'"

**Result:** The AI will respond as if it is directly helping the customer.

### 3. Empathy in First Person 

**Prompt:** "Use first person to show empathy. For example, instead of 'The system understands,' say 'I understand your concern.'"

**Result:** The AI shows a more empathetic and human-like response.

Using first-person language helps create a more conversational and relatable experience for your customers. It makes the AI feel more like a personal assistant, rather than a generic system, improving customer satisfaction.


To ensure the AI uses first-person responses, you can write custom prompts that guide the AI to answer in the first person. This makes interactions feel more personal and engaging.

Teaching AskHandle AI to Use First Person


# How to update payment information?

To add or change your payment information, navigate to the Settings section in your user dashboard. You can find Settings in the bottom-left corner of the menu, represented by a gear icon.
 
Once in the Settings section, locate the "Billing Management" subsection. Click on the "Add payment info" button to input new payment information, or click the “Update information” button to modify existing details, and fill out the forms with the updated information.
 
The navigation path to add new payment information is as follows: User dashboard -> Settings -> Billing Management -> Add payment info.
 
To update existing payment information, follow this navigation path: User dashboard -> Settings -> Billing Management -> Update information.

To add or change payment information, go to the Settings section in your user dashboard, accessible via the gear icon in the bottom-left corner. Within Settings, find Billing Management, where you can either add new payment details by clicking Add payment info or update existing information by selecting Update information and filling out the necessary forms. The navigation path for adding payment information is User dashboard -> Settings -> Billing Management -> Add payment info, while for updating existing details, it's User dashboard -> Settings -> Billing Management -> Update information.

How to update payment information?


# How to Use Keyword Notification?

The Keyword Notification feature is available for users on the Professional Plan and above. This feature allows you to set specific keywords in the Command Center, and the AI will send you a notification whenever a user mentions these keywords in a conversation.

## Setting Up Keywords

1. **Enter Keywords in the Command Center**: In the "Notification by Keyword Phrases" section of the Command Center, enter your desired keywords. You can add multiple keywords, separating each with a comma. Be sure to save your changes so the AI knows which keywords to monitor.
  
2. **Add Your Email**: Go to [Settings](https://dashboard.askhandle.com/users/~redirect/) and add your email under the "I'd like to receive an email notification for every inquiry" section. This ensures you’ll receive email notifications when the AI detects your specified keywords in a conversation.

## When Keywords Will Trigger Notifications

The keyword notification activates whenever a user mentions one of your specified keywords during a conversation. For example, if you’ve set "`price`" or "`pricing`" as keywords, you’ll receive an email notification if the user asks, "`What's your pricing?`" or "`What is the price?`"

Please note, however, that keyword notifications are based on user input only—they won’t trigger from the AI’s responses.

### Expanding the Use Case

You can also use keyword notifications for specific processes, such as scheduling. For instance, you might set the keyword to "`I confirm`" and instruct users to confirm appointments by typing this phrase. This approach allows you to expand the practical applications of keyword notifications for a range of scenarios.

The Keyword Notification feature is available for users on the Professional Plan and above. This feature allows you to set specific keywords in the Command Center, and the AI will send you a notification whenever a user mentions these keywords in a conversation.

How to Use Keyword Notification?


# Chat Widget Not Loading? Cookiebot or CSP May Be Blocking It

If you've correctly installed the chat widget using Google Tag Manager (GTM) or directly in a single-page React app — but it **still doesn’t appear** — the problem is likely due to:

* **Cookiebot blocking scripts until consent is given**
* **Content Security Policy (CSP) preventing external scripts**
* **Incorrect GTM tag configuration or load timing**

This article explains the cause and gives step-by-step solutions.


## 🚫 Common Errors You Might See

When inspecting your browser's developer console, you might see:

### 🔒 Cookiebot Blocking Scripts

```
consent.cookiebot.com/:1 Refused to frame 'https://consentcdn.cookiebot.com/' ...
```

This means **Cookiebot is blocking third-party scripts** (including your chat widget) before the user gives consent.

---

### 🔐 Content Security Policy (CSP) Errors

```
Refused to load the stylesheet 'https://www.googletagmanager.com/debug/badge.css' ...
```

This indicates your CSP is **too strict** and blocks resources like:

* Chat widget scripts
* GTM scripts
* External iframes or fonts

---

## ✅ How to Fix It

### 🔧 Fix 1: Allow Chat Widget Domain in Your CSP

Ask your developer or hosting team to add the following to your `Content-Security-Policy` (CSP) headers:

```http
script-src https://handle-chat-widget.s3.amazonaws.com;
frame-src https://handle-chat-widget.s3.amazonaws.com;
```

If you're using `'self'` in your CSP, you **must explicitly list** all allowed domains — including the chat widget script host.

---

### 🔧 Fix 2: Configure Cookiebot to Allow the Widget

Cookiebot blocks scripts until users consent to specific categories. Here's how to ensure it allows the chat widget:

#### Option A: Use `data-cookieconsent` Attributes

If injecting manually (HTML or GTM):

```html
<script type="text/plain" data-cookieconsent="functionality">
  window.webchatConfig = { token: "K3622313-109-356-0797850F" };
</script>

<script type="text/plain" data-cookieconsent="functionality" src="https://handle-chat-widget.s3.amazonaws.com/assets/js/webchat-widget.min.js"></script>
```

> This tells Cookiebot: “Only run this script if the user has given functionality consent.”

#### Option B: Set It Up in Cookiebot Dashboard

1. Log into [Cookiebot](https://manage.cookiebot.com/).
2. Go to **"Your Scripts"** or **"Custom Scripts"**.
3. Add a new script:

   * Set **Category** to `Functionality` or similar
   * Paste your chat widget script
   * Choose **Load after consent**

---

### 🔧 Fix 3: Configure GTM Tag Consent Settings

If you're using **GTM with Consent Mode**, configure your tag:

1. In GTM, edit the chat widget tag.
2. Go to **"Consent Settings"**.
3. Enable **“Require additional consent”**.
4. Add `functionality` (or whatever category you used in Cookiebot).

This ensures GTM will **wait for proper consent** before injecting the widget.

## Test Your Setup

After making these changes:

1. Clear cookies in your browser (to reset consent).
2. Visit your website.
3. Decline cookies — the widget **should not appear**.
4. Accept cookies — the widget **should now appear**.
5. In the Dev Console, check:

   * Is `window.webchatConfig` defined?
   * Did the script from `handle-chat-widget.s3.amazonaws.com` load?
   * Is the widget DOM element present?

## Summary

| Problem                       | Cause                          | Solution                                               |
| ----------------------------- | ------------------------------ | ------------------------------------------------------ |
| Widget not loading            | Cookiebot blocks scripts       | Use `data-cookieconsent` or configure in Cookiebot     |
| Widget script blocked         | CSP blocks third-party sources | Add `script-src` and `frame-src` for the widget domain |
| Widget not appearing via GTM  | Tag loads before consent       | Use Consent Settings and DOM Ready trigger             |
| Widget appears inconsistently | GTM/React race conditions      | Load the script dynamically with proper delay          |


If you've correctly installed the chat widget using Google Tag Manager (GTM) or directly in a single-page React app — but it **still doesn’t appear** — the problem is likely due to Cookiebot blocking scripts until consent is given, Content Security Policy (CSP) preventing external scripts, or Incorrect GTM tag configuration or load timing, this article explains the cause and gives step-by-step solutions.

Chat Widget Not Loading? Cookiebot or CSP May Be Blocking It


# Is My Imported Data Being Used by Anyone Else?

At AskHandle, your data is **entirely your own**. The information you import into your data source is used **exclusively** for your account and is never shared or utilized in any AI training outside of your system.

We take data privacy seriously. As pioneers of the codeless RAG system, we ensure that any content you import is used **only** to enhance your AI’s responses for your organization, services, or products. The AI leverages your data solely to answer queries from your users—no one else has access to it, and it is not incorporated into any general AI models or external systems.

You retain **complete control** over your data. Once you remove information from your data source, it is no longer accessible by the AI, and no remnants of it will remain. This gives you full authority over what the AI can and cannot use to provide customer support.

Simply put, **you own your training data and customer data**. It is not used for any purpose beyond serving your account, ensuring that your information remains private, secure, and under your control at all times.

At AskHandle, your data is entirely your own. The information you import into your data source is used **exclusively** for your account and is never shared or utilized in any AI training outside of your system.

Is My Imported Data Being Used by Anyone Else?


# Prohibited Content for AI Training and Fine-Tuning

At AskHandle, we maintain strict guidelines to ensure that our AI systems are developed in a safe, ethical, and responsible manner. To protect the integrity of our platform and ensure AI safety, **we do not allow the upload of certain types of content for training or fine-tuning AI**.

## Do NOT Upload These Types of Content:

1. **Hate Speech or Discriminatory Content**
   - Content that promotes hate or discrimination based on race, gender, religion, sexual orientation, disability, or other characteristics.
   - Any material intended to harm or incite violence against individuals or groups.

2. **Sexual or Explicit Content**
   - Pornographic, sexually explicit, or obscene materials.
   - Any documents that include sexual content inappropriate for general audiences.

3. **Violent or Harmful Content**
   - Content that glorifies violence, promotes harm to individuals or groups, or includes graphic depictions of violence.
   - Documents encouraging self-harm, abuse, or dangerous behaviors.

4. **Illegal or Unethical Content**
   - Documents promoting illegal activities such as hacking, drug use, or criminal behavior.
   - Any material that encourages unethical practices, fraud, or deception.

## Why This Matters

Uploading content that violates these rules can lead to:
- AI malfunctions and poor performance.
- Violation of AI safety policies, which could result in restricted access to AskHandle services.
- Harmful, biased, or inappropriate AI-generated content.

## Other Prohibited Use Cases
- Creating bots for harassment or cyberbullying.
- Training AI for spam or spreading misinformation.
- Generating offensive or inappropriate content that could harm or offend others.

Uploading these types of content undermines AI safety and violates AskHandle's ethical standards. Thank you for helping us maintain a safe and ethical AI environment.


At Askhandle, we maintain strict guidelines to ensure that our AI systems are developed in a safe, ethical, and responsible manner. To protect the integrity of our platform and ensure AI safety, we do not allow the upload of certain types of content for training or fine-tuning AI.

Prohibited Content for AI Training and Fine-Tuning


# Can I Use Images or Videos for Fine-Tuning?

AskHandle’s AI is built around **text-based training and fine-tuning**, meaning that only text can be used to set up and train your AI. If you try to import content that includes images or videos—such as a webpage with visual elements or files that embed multimedia content—the AI will not be able to process that information. Fine-tuning can only occur when the imported data is purely text-based.

### Why Must Fine-Tuning Be Text-Based?

AskHandle is designed to learn from **text** to generate responses. Any content that is not text, like images or videos, cannot be used in the fine-tuning process. Even if you have text embedded within an image, such as on a webpage or promotional material, AskHandle will not be able to interpret that as usable data for fine-tuning.

### What Can You Do?

To properly fine-tune your AI, it's essential to **convert all information into text form**. Here are a couple of options:

- **Import Text Files:** If your images or videos contain important information, extract the text and upload it in text-based formats like PDF, DOCX, or plain text.
- **Manually Extract Text:** For content embedded in images or videos, manually transcribe the text and import it into the Command Center. This ensures the AI learns from the right information and can use it to generate accurate answers.

Once the text is imported, AskHandle can fine-tune its responses and use that information to provide targeted, context-aware answers.

### Can AI Include Images in Responses?

While AI fine-tuning is based on text, you can still [teach your AI to include images in its responses](/how-to-include-pictures-into-ai-responses). Simply add instructions or links in the text-based content that direct the AI to use specific images when providing answers. This allows you to incorporate visual elements into the chatbot's replies without using images as part of the training process.

By focusing on text for fine-tuning and incorporating images through instructions, you can create a more engaging and responsive AI experience.

AskHandle’s AI is built around text-based training and fine-tuning, meaning that only text can be used to set up and train your AI. If you try to import content that includes images or videos—such as a webpage with visual elements or files that embed multimedia content—the AI will not be able to process that information. Fine-tuning can only occur when the imported data is purely text-based.

Can I use images or videos for fine-tuning?


# What is a Channel?

AskHandle is available on web messenger and WhatsApp. Every plan, by default, comes with a web messenger that can be easily installed on any website. Only paid plans are eligible for the WhatsApp channel. Please contact us directly at support@askhandle.com if you’d like to have AskHandle on WhatsApp.

If you would like to have AskHandle on any other communication channel besides web messenger and WhatsApp, including SMS or voice, please contact us at support@askhandle.com, and we will be more than happy to assist you.


AskHandle is available on web messenger and WhatsApp. Every plan, by default, comes with a web messenger that can be easily installed on any website. Only paid plans are eligible for the WhatsApp channel. Please contact us directly at support@askhandle.com if you’d like to have AskHandle on WhatsApp.

What is a Channel?


# What Are Live Chat Capabilities?

AskHandle is, first and foremost, an automated customer support system with the sole goal of eliminating any manual workload from the customer support services.
 
We understand that some individuals still like to have the option to jump into the conversation with users and chat by themselves. For this reason, we have added a live-chat capability.
 
In the Chats section, you can see previous and current chats. If the chat is ongoing and you’d like to join, simply turn off the AI and chat. There’s a toggle named “Use AI”; turn it off and use the text field below the current chat to write your responses, then click on “Send message” whenever you want to send your message to the user.
 
When the AI is turned off, AskHandle becomes a typical live-chat solution.



AskHandle primarily serves as an automated customer support system, but also offers a live-chat feature for users who prefer human interaction, accessible by toggling off the AI in the Chats section.

What Are Live Chat Capabilities?


# What are Messages?

 This is the part where the entire interaction between users and the chatbot is stored. This is also the part where you can directly chat with users, if you turned off the automation.

To access this section, please log in to your account and click on the “Messages” button in the “Menu”.

All previous interactions and current interactions can be seen in the “Messages” section. On the right side, you have the option to quickly find any previous interaction by typing the user’s name in the search bar.

Type in the name of the user and the interaction with that user’s name will be shown in the results.

Turn off the AI
If you or your customer representatives are available and want to take over the interaction, you can turn off the AI. This will disable smart automation (no more automatic answers) and your representatives can chat directly with the users.

Option 1: Turn off AI and join a conversation.
Hover over the left sidebar menu, go to the “Messages” section, and click on the “Use AI” button; AI will be disabled for that particular chat. If you’d like to turn it back on, click on the “Use AI” button again and if there’s a blue color, the chatbot is on again.

Option 2: Turn off AI globally.
You can turn off AI globally, not just for a particular chat but for every new chat you receive, so AskHandle becomes a live-chat tool, and your representatives can chat with users.

If you’d like to completely turn off AI, hover over the left sidebar menu, go to “Appearance” -> Autopilot -> Use AI during the chat, and turn it off.



To access and manage user interactions, log in to your account, navigate to the Messages section in the menu, where you can review past and ongoing conversations, search for specific users, and toggle between automated and direct chat options. Additionally, you can globally disable AI to enable live-chat capabilities for representatives.

What are Messages?


# What Does 'Load URL Recursively' Mean?

The '**Load URL Recursively**' feature in AskHandle AI allows the system to learn not just from a single webpage URL you provide, but also from its child pages. This feature is particularly useful for creating a comprehensive data source from structured websites like help centers or FAQ sections.

![AskHandle data source mode](/images/help/articles-img/help-articles/load-url-recursively.png)

## How 'Load URL Recursively' Works

When you enable the **'Load URL Recursively'** option, the AI will follow links from the specified parent page to its child pages and gather content from them. 

**Example:**
If you input the URL:  
`https://www.example.com/help`  

And this page contains links like:  
- `https://www.example.com/help/how-to-reset-password`  
- `https://www.example.com/help/how-to-contact-customer-support`  
- `https://www.example.com/help/how-to-get-refund`  

The AI will visit and learn from these linked pages as well. Currently, AskHandle AI supports learning up to **3 levels of links**:

1. **Parent Page**: The page you input.
2. **Direct Child Pages**: Links directly found on the parent page.
3. **Child Pages of Each Child Page**: Links found on each direct child page.

This 3-level structure is designed to cover the majority of help center and FAQ content effectively.

If your goal is to train the AI on your help center or FAQ section, simply input the parent URL, and the AI will automatically process the relevant linked pages.

## Limitations to Keep in Mind

1. **Website Restrictions:** Some websites, particularly those with anti-scraping measures or firewalls like CAPTCHA, may block the AI from accessing their content. In such cases, the AI will only process the original URL you provide.
2. **Large Parent Sites:** If the parent page has an extensive number of child pages, the AI might not capture all of them in a single attempt. To address this, you can:
   - Add specific subpages individually as additional data sources.
   - Contact our customer support team at [support@askhandle.com](mailto:support@askhandle.com) for further assistance.

## Best Practices

- Use the 'Load URL Recursively' option for websites you manage or own, ensuring no restrictions prevent the AI from accessing the content.
- If you need finer control over the AI's learning process, you can input individual URLs as separate data sources.

The 'Load URL Recursively' feature simplifies the process of training AskHandle AI on large, structured websites, making it easier to create a robust and informative AI response system.


The 'Load URL Recursively' feature in AskHandle AI allows the system to learn not just from a single webpage URL you provide, but also from its child pages. This feature is particularly useful for creating a comprehensive data source from structured websites like help centers or FAQ sections.

What Does 'Load URL Recursively' Mean?


# What is a Data Source?

A **data source** is a repository where your information is stored, and it serves as the foundation for AskHandle to generate answers and content. You can import data into a data source using three methods: uploading files (PDF, DOCX, CSV, TXT, or Markdown), inserting a webpage URL, or directly typing or pasting content into a plain text field. You can use one, two, or all three methods depending on your needs.

### What Kind of Data Should I Include? Do I Need to Add Questions?

The data you import should be relevant to the goals you want your AI to achieve. Any information you want AskHandle to provide as an answer to user queries should be included in your data source. You don't need to include specific questions—AskHandle is pre-trained to understand user queries, regardless of how they are phrased or the language used. For instance, you can upload a PDF containing details about your services, product features, or contact information. The AI will automatically learn from this document, and when a user asks a related question, it will respond with the appropriate information from your uploaded content.

### Can I Use a Webpage URL to Train the AI?

Yes, you can train your AI using a webpage URL. This is one of the three ways to import data into a data source and is a simple method to provide relevant information. However, keep in mind that if the webpage you choose doesn’t allow crawling, AskHandle won't be able to access its content. In such cases, the AI won't be able to learn from the webpage, and it won't be able to answer queries based on that data. It's important to ensure the webpage permits crawling when using this method.

### Is the AI Ready to Use Once I Import Data?

After you import and save your data into the data source, give the AI a few seconds to process the information. Refresh the Command Center page, and when the “Processed” label appears next to your data source, it means your chatbot is ready. If the label still says “Unprocessed,” wait a few more seconds and refresh the page again until the status changes to “Processed.” This indicates that your chatbot has successfully processed the data and is ready to answer user queries based on the information you've provided.

A data source is a repository where your information is stored, and it serves as the foundation for AskHandle to generate answers and content. You can import data into a data source using three methods: uploading files (PDF, DOCX, CSV, TXT, or Markdown), inserting a webpage URL, or directly typing or pasting content into a plain text field. You can use one, two, or all three methods depending on your needs.

What is a Data Source?


# What Is the AI Answer Relevance Score?

At AskHandle, we leverage AI to not only answer customer inquiries but to ensure those responses are directly relevant. The **AI Answer Relevance Score** plays a key role in evaluating how closely the AI’s responses address a user’s question, helping to refine the quality of answers. This score, which typically ranges from 0 to 1, provides a measure of how well the AI response aligns with the question’s intent, based on available data.

### What is the AI Answer Relevance Score?

The **AI Answer Relevance Score** measures how closely the AI’s answer matches the user’s question, focusing on relevance rather than factual accuracy. A high score means the response is highly relevant, while a lower score indicates that the answer may be incomplete, contain unnecessary details, or veer off-topic.

### How Does the Answer Relevance Score Work?

To determine answer relevance, AskHandle’s system follows a unique process:

1. **Reverse-Engineering Questions**: When the AI generates an answer, it creates several alternate questions based on that response, using a large language model (LLM). This process helps simulate whether the answer alone could lead back to the original question.
  
2. **Measuring Similarity**: The system then calculates the average cosine similarity between these generated questions and the original user question. A higher similarity score means the answer more directly addresses the question.

For example:

- **Question**: “How can I reset my account password, and are there any restrictions on password length?”
    - **Low Relevance Answer**: “To reset your password, go to the account settings page.”
    - **High Relevance Answer**: “To reset your password, go to the account settings page. Your new password must be at least 8 characters long and include both letters and numbers.”

### What Do the Scores Mean?

- **Score Close to 1**: Indicates a highly relevant response that directly addresses the question.
- **Score Near 0**: Shows the answer lacks relevance, either missing key information or including unrelated details.
- **Scores Between 0 and 1**: Reflect varying levels of relevance—partially related but not fully addressing the query.

The AI Answer Relevance Score is a key tool for evaluating how well the AI responds to customer questions. It provides clear insights into how accurately the AI uses available information, enabling businesses to fine-tune their systems and enhance the customer experience over time. Monitoring and acting on these scores helps companies ensure their AI consistently delivers useful and relevant responses.

At AskHandle, we leverage AI to not only answer customer inquiries but to ensure those responses are directly relevant. The AI Answer Relevance Score plays a key role in evaluating how closely the AI’s responses address a user’s question, helping to refine the quality of answers. This score, which typically ranges from 0 to 1, provides a measure of how well the AI response aligns with the question’s intent, based on available data.

What Is the AI Answer Relevance Score?


# What is a chat interaction?

A chat interaction refers to the activity where an individual user initiates a chat, engages in a conversation with AskHandle, and then leaves the chat. This interaction encompasses the entire conversation from start to finish, regardless of the number of messages exchanged within the chat session.

There is no limitation on the number of messages a user can send during the interaction, and users are not charged per message. Chat interactions are calculated by individual users who initiate a chat, engage in a conversation with AskHandle, and then leave the chat.

We do not limit the number of messages within the chat, nor do we charge per message.

A chat interaction comprises the entire conversation between a user and AskHandle, starting from initiation to conclusion, regardless of message count. Users aren't restricted in message quantity or charged per message, and interactions are calculated per individual user who initiates and concludes a chat session.

What is a chat interaction?


# What Is Command Center?

The **Command Center** is your hub for managing and customizing the information that powers your AI. It allows you to fine-tune your AI, customize interactive options, and manage responses. Through the Command Center, you can ensure that AskHandle performs optimally for your specific business needs.

## How to Access the Command Center

To access the Command Center, log in to your AskHandle account at [Command Center](https://dashboard.askhandle.com/stories/command_center/), hover over the left sidebar menu, and select "Command Center." 

## Training Your AI with Specific Information

Training your AI in the Command Center is a straightforward process. The way your AI responds to queries depends largely on the information you provide and how it is structured. This training phase is called "Fine-Tuning" and ensures your AI delivers relevant, accurate, and branded responses.

## Step-by-Step Guide to Using the Command Center

1. **Go to Command Center**

   Log in to your AskHandle account and navigate to the Command Center.

2. **Select the Data Source Mode**

   Choosing the right data source mode is essential to ensure your AI delivers responses that match your requirements. There are three data source modes available:

   - **General Data Mode**  
     AskHandle generates responses using publicly available information, primarily from the internet. This mode is suitable for common questions and general knowledge, but does not rely on any customized data you’ve uploaded.
  
   - **Your Own Data Mode**  
     In this mode, the AI uses only the data you’ve uploaded to generate responses. It won't access public information, ensuring that your responses are highly tailored and specific to your business.
  
   - **Combined Data Mode**  
     Combined Data Mode prioritizes your uploaded data while also using publicly available information when needed. This mode ensures personalized responses and allows small talk, such as casual greetings.

   For most cases, we recommend **Combined Data Mode** for optimal AI performance.

3. **Add a Data Source**

   After selecting your data mode, click the "Add Data Source" button to start training your AI. You can input your data in one of three ways:
   
   - **Document Files:** Upload DOCX, PDF, CSV, or TXT files.
   - **Plain Text:** Paste or write your data directly into the text field.
   - **Webpage URL:** Let the AI learn from a specific webpage (ensure you own or manage the content).

4. **Save the Data Source**

   Once you've added your data, click the "Save Data Source" button.

5. **Monitor Processing Status**

   After saving, the AI will process your input. This typically takes 10-20 seconds. Refresh the page to check the status, which will update from "UNPROCESSED" to "PROCESSED" once the AI has learned from the data.

## Setting Up Keyword Notifications

You can set up keyword notifications to monitor specific customer interactions. When a customer mentions the predefined keywords, the AI will send you an email notification. This ensures you're aware of important conversations in real time.

## Configuring Interactive Options

Interactive options allow users to click buttons for instant AI-powered replies. These dynamic responses are based on your fine-tuned data. To configure them:

- **Edit Interactive Options:** Click the "Edit" button next to the option, modify the text, and save your changes.
  
- **Remove Interactive Options:** Uncheck the "Show Button" box in the edit mode, and save the changes to remove the option from view.

Interactive options enhance the user experience by offering quick replies, so think carefully before editing or removing them.


The Command Center is your hub for managing and customizing the information that powers your AI. It allows you to fine-tune your AI, customize interactive options, and manage responses. Through the Command Center, you can ensure that AskHandle performs optimally for your specific business needs.

What Is Command Center?


# What is the Faithfulness Score in AI Responses?

The Faithfulness Score helps measure how accurately an AI-generated response aligns with the specific information provided by your business. It focuses on factual consistency, ensuring that the AI’s answers reflect the data you’ve supplied without adding or changing key details.

## What Does the Faithfulness Score Mean?

The Faithfulness Score, ranging from 0 to 1, shows how well the AI’s response matches the facts from your data:

- **Score of 1**: A fully accurate answer where all statements in the response can be verified by your data.
- **Score of 0**: An unfaithful answer where none of the information aligns with the provided context.
- **Scores between 0 and 1**: Partial accuracy, where some statements are supported by the data but others may not be.

A high Faithfulness Score indicates that every claim made in the AI’s answer can be inferred directly from the data source, while a lower score means the response may contain inaccuracies or assumptions that aren’t supported by the data.

## How is the Faithfulness Score Calculated?

To calculate the Faithfulness Score, the system follows these steps:

1. **Identify Claims in the Answer**: The AI breaks down the response into individual claims or statements.
2. **Verify Each Claim**: Each claim is then checked against the data provided to see if it is fully supported by the context.
3. **Calculate the Score**: Based on the percentage of claims that align with the provided context, a Faithfulness Score between 0 and 1 is assigned.

### Example

- **Question**: "What are the hours for customer service, and is there a weekend helpline?"
  
    - **Context**: "Our customer service team is available Monday through Friday, 9 AM to 6 PM. Weekend support is not available."
  
    - **High Faithfulness Answer**: "Customer service is available Monday to Friday, 9 AM to 6 PM, with no weekend support."
  
    - **Low Faithfulness Answer**: "Customer service is available daily, including weekends, from 9 AM to 6 PM."

In this example, the high-faithfulness answer matches the provided context exactly, while the low-faithfulness answer introduces incorrect information about weekend support.

## Why is the Faithfulness Score Important?

The Faithfulness Score ensures that the AI is providing factually consistent information based on your data, which is crucial in areas where accuracy matters, such as customer support, technical help, healthcare, or legal services. A high score means the AI is sticking closely to your facts, reinforcing customer trust and minimizing misinformation.

A lower score can indicate that the AI is including creative or inaccurate details, which may not be ideal when customers require precise information.

## How Can You Use the Faithfulness Score?

Monitoring the Faithfulness Score helps ensure that your AI is providing the most accurate responses possible. If low scores appear frequently, it may suggest a need to adjust or expand your data to better cover the types of questions customers are asking.

The Faithfulness Score is a valuable tool for tracking the reliability of the AI’s answers, ensuring that your customers receive factually correct and relevant information every time they interact with your AI. This focus on accuracy helps build trust and improves the overall quality of your customer support.

The Faithfulness Score helps measure how accurately an AI-generated response aligns with the specific information provided by your business. It focuses on factual consistency, ensuring that the AI’s answers reflect the data you’ve supplied without adding or changing key details.

What is the Faithfulness Score in AI Responses?


# What is a Performance report?

If you opt in to receive performance reports, which are enabled by default, you will receive a report every day. This report will detail the questions your users had, categorized into specific topics for easier management.
 
The reports provide an overview of the information your users inquire about and highlight the most commonly triggered topics.
 
If you prefer not to receive daily reports, you can easily opt out of them in the Settings section.


Opting in to receive daily performance reports from AskHandle offers a categorized overview of user questions, facilitating easier management, though opting out is simple in the Settings section.

What is a Performance report?


# What is the Testing Page in the Dashboard?

The **Testing Page** in your dashboard is a specialized page designed to help you thoroughly test your AI model and customize the appearance of the chat interface. It provides a real-time preview environment, allowing you to review and refine how your AI interacts with users before you launch it publicly. You can easily access the Testing Page through a unique URL or QR code.

To start testing your AI, visit your [Testing Page](https://dashboard.askhandle.com/landing-page/). Before you begin, make sure to assign a clear name to your Testing Page URL and save the settings. Once saved, you'll receive a unique URL specifically for your testing needs.

## Key Features of the Testing Page:

- **Real-time AI Testing:**  
  Interact directly with your AI to evaluate its conversational flow, responsiveness, and accuracy before deploying it to users.

- **Customizable Appearance:**  
  Tailor the Testing Page to align with your brand identity. You can add a custom background image and your own logos, creating a consistent, branded testing experience.

- **QR Code Generation and Customization:**  
  Easily generate a customized QR code featuring your logo. This makes it convenient to test your AI on mobile devices by simply scanning the code.

- **Mobile-Friendly Testing:**  
  Scan your QR code with a mobile device to instantly access the Testing Page. This allows you to evaluate the AI’s performance and appearance across different devices seamlessly.

The Testing Page ensures that your AI not only performs well but also visually aligns with your branding and provides users with a smooth, engaging experience.

The Testing Page in your dashboard is a specialized page designed to help you thoroughly test your AI model and customize the appearance of the chat interface. It provides a real-time preview environment, allowing you to review and refine how your AI interacts with users before you launch it publicly. You can easily access the Testing Page through a unique URL or QR code.

What is the Testing Page in the Dashboard?


# What Is the User Dashboard?

The User Dashboard serves as the central hub for managing your AI, messenger services, and conversations. Accessible at [AskHandle Dashboard](https://dashboard.askhandle.com/accounts/login/), this platform provides comprehensive tools for optimizing AI performance and user interactions.

## Key Features and Capabilities

1. **AI Customization and Training**  
   Tailor your AI by integrating new data sources and refining its training. This ensures that your AI is well-equipped to address customer inquiries with precision and relevance.

2. **Performance Monitoring**  
   Track real-time performance metrics, including response accuracy and user engagement. Access detailed analytics to evaluate chat histories and ensure consistent quality in customer interactions.

3. **Customer Insights**  
   Gain in-depth understanding of customer behavior and profiles. The dashboard enables you to monitor interactions, providing valuable insights into your audience's needs and preferences.

4. **Visual and UX Customization**  
   Adjust the look and feel of the chat interface to align with your brand’s design. Modify elements of the messenger’s appearance and user experience to enhance customer satisfaction.

5. **Account Management**  
   Manage notifications, account settings, and subscription details, including billing preferences. Easily stay up to date with your account status and control important settings.

Sign in to the [AskHandle Dashboard](https://dashboard.askhandle.com/accounts/login/) to take full advantage of these powerful tools and optimize your AI’s capabilities.


The User Dashboard serves as the central hub for managing your AI, messenger services, and conversations. Accessible at AskHandle Dashboard, this platform provides comprehensive tools for optimizing AI performance and user interactions.

What Is User Dashboard?

# When should I use the REST API-based chat widget?

AskHandle offers two types of chat widgets:

* A **WebSocket-based widget** for real-time, live human chat
* A **REST API-based widget** for lightweight, customizable, AI-powered chat

This article explains **when to use the REST API widget**, how it differs from the default, and how to install and deploy it.

## ✅ Use the REST API Widget If:

* You're building an **AI chatbot experience** (no live agents)
* You want a **lightweight, customizable widget**
* You need **complete control** over hosting, style, and behavior
* You're working with **static sites**, **headless CMS**, or **JAMstack**
* You want to manage deployment for **performance or branding**

## WebSocket vs. REST: Which Widget Should I Use?

| Feature                         | WebSocket-Based Widget (Default)     | REST API-Based Widget (Open Source)   |
| ------------------------------- | ------------------------------------ | ------------------------------------- |
| **Live chat with human agents** | ✅ Yes – real-time support            | ❌ No – not for live human chat        |
| **AI chatbot support**          | ✅ Yes                                | ✅ Yes – ideal for async AI responses  |
| **Customization**               | Limited (via dashboard)              | Full (via JS config file)             |
| **Installation**                | One-line embed (AskHandle dashboard) | One-line embed (from your own server) |
| **Hosting**                     | Hosted by AskHandle CDN              | You host it yourself                  |
| **Performance**                 | Heavier (persistent WebSocket)       | Lightweight (HTTP only)               |

> **Use the WebSocket widget** if you're offering live chat with real human agents.

> **Use the REST API widget** if you're building an AI-powered assistant that doesn't require real-time messaging.

## How to Install and Deploy the REST API-Based Chat Widget

This step-by-step guide walks you through everything needed to:

1. **Download the open-source widget**
2. **Customize it**
3. **Deploy it using Heroku, Render, or any Node.js host**
4. **Embed it in your website**

### Step 1: Download the Chat Widget Code

Clone the official GitHub repository:

```bash
git clone https://github.com/Forte-AI/light-chat-widget.git
cd light-chat-widget
```

---

### Step 2: Install Dependencies

Make sure you have **Node.js v14 or higher**:

```bash
npm install
```

---

### Step 3: Set Up Your API Token

Create a `.env` file in the project root with your AskHandle API token:

```env
ASKHANDLE_API_TOKEN=your-api-token-here
```

You can find this in the **AskHandle dashboard > Settings > API Access**.

---

### Step 4: Customize the Widget

Open `static/widget-config.js` to configure layout, position, and dimensions:

```javascript
const config = {
  widgetUrl: 'https://your-deployment-url.com/widget',
  position: 'bottom-right',
  iconPosition: 'bottom-right',
  width: '370px',
  height: '650px',
  verticalOffset: '20px',
  horizontalOffset: '20px'
};
```

---

### Step 5: Deploy to a Hosting Platform

#### 🚀 Option A: Deploy to Heroku

```bash
heroku create your-widget-app
heroku config:set ASKHANDLE_API_TOKEN=your-api-token-here
git push heroku main
```

#### 🚀 Option B: Deploy to Render

1. Go to [Render.com](https://render.com/) and sign up.
2. Create a **new Web Service**.
3. Connect your GitHub repo.
4. Set `ASKHANDLE_API_TOKEN` as an environment variable.
5. Use `npm start` as the start command.

---

### Step 6: Embed the Widget on Your Website

Once deployed, copy your hosted `widget-config.js` URL and paste it just before the `</body>` tag on your site:

```html
<script src="https://your-widget-app.onrender.com/widget-config.js"></script>
```

---

### Example: Full Embed Code

```html
<!DOCTYPE html>
<html lang="en">
<head>
  <meta charset="UTF-8">
  <title>My Website</title>
</head>
<body>

  

  
  <script src="https://your-widget-app.onrender.com/widget-config.js"></script>
</body>
</html>
```

---

## 🧼 Optional: Local Development Commands

| Task                    | Command         |
| ----------------------- | --------------- |
| Start dev server        | `npm run dev`   |
| Build for production    | `npm run build` |
| Start production server | `npm start`     |

---

## 🔗 When Should I Use the Default Widget Instead?

Use the **WebSocket-based widget** if you:

* Need **live chat with human agents**
* Want the **fastest plug-and-play** install via the AskHandle dashboard
* Don’t want to host the widget yourself

**Default widget install code:**

```html
<script>
  window.webchatConfig = { token: "G3623219-369-356-1797850F" };
</script>
<script src="https://handle-chat-widget.s3.amazonaws.com/assets/js/webchat-widget.min.js"></script>
```

* Open-source widget repo: [GitHub – Forte-AI/light-chat-widget](https://github.com/Forte-AI/light-chat-widget)

AskHandle offers two types of chat widgets: A WebSocket-based widget for real-time, live human chat, a REST API-based widget for lightweight, customizable, AI-powered chat. This article explains when to use the REST API widget, how it differs from the default, and how to install and deploy it.

When should I use the REST API-based chat widget?


# When Will the User See the Contact Message?

When you sign up and begin a test chat on the Testing Page, you may notice that the AI repeatedly responds with the same message. This response is known as the *Contact Message*, which you can customize in the *Appearance*.

### Why Does the AI Respond with the Contact Message?

The Contact Message appears when you start using the AI without adding any data sources. By default, the AI operates in *General Data* mode, which relies on both built-in knowledge and user-provided data. If no data source has been added, the AI defaults to responding with your set Contact Message as it has no other content to draw upon.

### How to Enable More Dynamic Responses

For the AI to provide more tailored and logical responses—including engaging in small talk—you need to add at least one data source. This ensures that the Combined Mode has the necessary information to function effectively.

### Alternative Option: General Data Mode

If you prefer not to add a data source but still want the AI to provide relevant responses, switch to *General Data Mode*. In this mode, the AI responds based on its own knowledge, bypassing the need for custom data sources.

### Customizing the Contact Message

You can set up your own Contact Message in the [Appearance](https://dashboard.askhandle.com/customization). This message will also be used as a fallback response in *Your Own Data Mode* when the AI cannot find relevant information in your data source. 


When you sign up and begin a test chat on the Testing Page, you may notice that the AI repeatedly responds with the same message. This response is known as the Contact Message, which you can customize in the *Appearance* settings.

When Will the User See the Contact Message?


# Where Do We Store Your Data at AskHandle?

At AskHandle, data storage is a top priority, and we use secure, high-performance cloud services to keep your data safe, accessible, and well-managed. Our database infrastructure is now powered by **Heroku Postgres**, a reliable and scalable SQL database solution directly managed by Heroku. This transition allows us to offer improved flexibility, security, and performance for our users.

## Database Location and Compliance

Our Heroku Postgres database is hosted on servers in the United States, aligning with our main application server location. This setup provides optimal performance and compliance with U.S.-based regulations, such as the **California Consumer Privacy Act (CCPA)** and **Health Insurance Portability and Accountability Act (HIPAA)**. Data stored on U.S. servers adheres to local data protection standards, ensuring compliance with American privacy laws.

## What PostgreSQL Version Are We Using?

With Heroku Postgres, we have access to stable, recent versions of PostgreSQL. Heroku frequently updates its PostgreSQL instances to provide the latest features, performance enhancements, and security improvements. This ensures our database infrastructure remains current and fully functional, supporting the best possible experience for our users.

## Heroku Postgres Features and Dashboard

Heroku Postgres offers a powerful web dashboard for easy database management. You can monitor performance, view usage statistics, and even share queries through *dataclips*. Additionally, you can manage the database using Heroku CLI commands, giving flexibility in how the database is maintained.

### Choosing the Right Heroku Postgres Plan

Heroku Postgres provides a range of service plans, including Essential, Standard, Premium, Private, and Shield. Each plan has different resources, so as your data storage needs grow, it’s easy to upgrade to a more robust option.

## High-Performance Infrastructure

Heroku’s infrastructure ensures our databases run efficiently, with performance optimized for high-demand applications. Each database instance is hosted on dedicated, isolated virtual machines within Heroku’s cloud environment. This setup guarantees the speed, reliability, and isolation necessary to handle concurrent connections and large volumes of queries smoothly.

## Security and Data Protection

Heroku Postgres includes several robust security features, providing peace of mind for users:

- **IP Restrictions**: Heroku allows us to control database access by IP, ensuring only authorized users can connect.
- **No Direct SSH Access**: Heroku manages the underlying infrastructure, applying regular updates and patches to maintain security, which eliminates the need for SSH access.
- **Backups and Disaster Recovery**: Heroku Postgres includes daily backups, stored within the U.S., to protect data against loss. For additional disaster recovery, continuous protection features are available to capture snapshots and transaction logs in the same region as the database.

## Local Development and Environment Parity

To ensure smooth deployment, Heroku recommends setting up Postgres locally, matching your local development environment to the production database. This simplifies testing and deployment by keeping the development and production databases consistent.

## Managing Multiple Applications

Heroku allows us to share a single Postgres database across multiple applications using the `heroku addons:attach` command. This is helpful for apps that need to collaborate on data but still operate independently. Config vars allow us to assign aliases to shared databases, making management straightforward and organized.

## Logging and Privacy Options

For additional privacy, Heroku provides options to block logging of database queries. This prevents sensitive data from appearing in logs, although it may limit some debugging capabilities. For detailed logging needs, we recommend selectively logging specific data points.

## Flexible Plan Upgrades

Heroku Postgres allows easy migration between plans as application needs evolve. When our app’s requirements exceed the resources of the current plan, upgrading to a larger plan is straightforward, ensuring our database can scale with our users’ demands.

## Consistent, Reliable Backups

Backups are stored within the U.S. and managed by Heroku’s continuous protection systems. Snapshots and transaction logs are kept consistent, ensuring data integrity and enabling us to restore data effectively if necessary. This approach ensures that all backups are reliable and encrypted, safeguarding data even in storage.

AskHandle’s integration with Heroku Postgres provides a secure, flexible, and high-performance data storage solution. Optimized for U.S.-based compliance, this setup ensures reliable and efficient access to data across our platform, meeting the highest standards of performance and security for our users.


At AskHandle, data storage is a top priority, and we use secure, high-performance cloud services to keep your data safe, accessible, and well-managed. Our database infrastructure is now powered by Heroku Postgres, a reliable and scalable SQL database solution directly managed by Heroku. This transition allows us to offer improved flexibility, security, and performance for our users.

AskHandle Help Center

Getting started

Featured articles

Can’t find what you are looking for?