export const article = {
  author: 'Samuel Su',
  date: '2025-11-25',
  title: `WhatsApp Web API: Send Outbound Messages`,
  label: 'feature',
  avatar: '/images/samuel_su_profile.png',
  hashtags: ['API', 'WhatsApp', 'Messaging'],
  thumbnail,
  preview: `Initiate conversations with your customers directly through WhatsApp Web API.

**What's New:**
- New POST **/send-whatsapp-web-message-outbound** endpoint to send outbound WhatsApp messages
- **File attachments support** - Send documents, images, and other files with your messages
- **Audio message support** - Convert text to natural-sounding audio messages
- Seamless conversation threading - replies from customers automatically continue in the same conversation
- Automatic lead creation and tracking for new contacts
- Perfect for customer notifications, proactive support, and marketing campaigns
`,
};


We're excited to announce the ability to send outbound messages through the WhatsApp Web API integration. You can now initiate conversations with your customers programmatically, enabling proactive customer engagement at scale.

---

## What's New

Send WhatsApp messages to your customers directly through our API, and when they reply, the conversation continues seamlessly in the same thread.

**Key Features:**

🚀 **Initiate Conversations** - Start WhatsApp conversations with customers programmatically

💬 **Automatic Threading** - Customer replies automatically attach to the same conversation

👥 **Lead Management** - New contacts are automatically created as leads in your system

🔗 **Conversation Tracking** - Each conversation gets a unique ID for easy reference and follow-up

📎 **File Attachments** - Send documents, images, PDFs and other files with your messages

🔊 **Audio Messages** - Convert text to natural-sounding audio for voice-based engagement

---

## Key Use Cases

**Customer Notifications**
- Order confirmations and shipping updates
- Appointment reminders and scheduling confirmations
- Account alerts and security notifications

**Proactive Support**
- Follow-up messages after purchases
- Check-ins with customers about their experience
- Proactive troubleshooting and assistance

**Marketing & Engagement**
- Personalized promotional campaigns
- Re-engagement messages for inactive customers
- Event invitations and announcements

---

## Message Formats & Attachments

**Text Messages (Default)**
Send standard text messages to your customers.

**File Attachments**
Include documents, images, PDFs and other files with your messages:
- Up to 3 files per message
- Supports images, PDFs, documents, and more
- Perfect for sending receipts, tickets, invoices, product images

**Audio Messages**
Enable audio format to convert your text into natural-sounding voice messages:
- Requires chatbot audio settings to be enabled
- Automatic text-to-speech conversion
- Ideal for voice-first customer experiences

---

## How It Works

When you send an outbound message through the API, the system:

1. Creates or updates the customer's lead profile automatically
2. Generates a unique conversation ID for threading
3. Delivers the message via your connected WhatsApp Web integration
4. Routes any replies from the customer back to the same conversation

This means you get complete conversation history and context, even when you initiate the conversation.

---

## Quick Examples

### Basic Text Message

```bash
POST /api/v2/send-whatsapp-web-message-outbound
Authorization: Bearer YOUR_API_KEY
Content-Type: application/json

{
  "chatbotId": "your_chatbot_id",
  "phone": "1234567890",
  "message": "Hi! Your order #12345 has been shipped and will arrive tomorrow."
}
```

### Message with File Attachment

```bash
POST /api/v2/send-whatsapp-web-message-outbound
Authorization: Bearer YOUR_API_KEY
Content-Type: application/json

{
  "chatbotId": "your_chatbot_id",
  "phone": "1234567890",
  "message": "Here's your invoice for order #12345",
  "files": [
    {
      "name": "invoice-12345.pdf",
      "type": "application/pdf",
      "url": "https://example.com/invoices/12345.pdf"
    }
  ]
}
```

### Audio Message

```bash
POST /api/v2/send-whatsapp-web-message-outbound
Authorization: Bearer YOUR_API_KEY
Content-Type: application/json

{
  "chatbotId": "your_chatbot_id",
  "phone": "1234567890",
  "message": "Hi! Your order #12345 has been shipped and will arrive tomorrow.",
  "format": "audio"
}
```

**Response (same for all formats):**

```json
{
  "status": "success",
  "conversationId": "1234567890:0987654321",
  "leadUUID": "1234567890"
}
```

When the customer replies to your message, their response will automatically appear in your Chat Data inbox under the same conversation ID.

---

## Getting Started

**Step 1:** Connect your WhatsApp Web account in Chatbot Settings → Third-Party Integrations

**Step 2:** Get your API key from Account Settings → API Keys

**Step 3:** Send your first outbound message using the endpoint above

**Step 4:** Monitor replies in your Chat Data inbox or via the conversation API

---

## Important Notes

- Phone numbers should be in E.164 format without the '+' prefix (e.g., "1234567890")
- Your WhatsApp Web integration must be connected and authenticated
- **Files**: Maximum 3 files per message, each file must have name, type (MIME type), and url properties
- **Audio**: Requires chatbot audio settings to be enabled in your chatbot configuration
- All replies from customers will appear in your Chat Data inbox
- Conversation history is preserved for both outbound and inbound messages

---

## Documentation

For complete API specifications, authentication details, and error handling, see:

- [API Reference - /send-whatsapp-web-message-outbound](/api-reference#tag/Chatbot-Operations/operation/SendWhatsAppWebMessageOutbound)

Available now for all plans with WhatsApp Web integration and API access.

WhatsApp Web API: Send Outbound Messages

export const article = {
  author: 'Samuel Su',
  date: '2025-11-24',
  title: `Claude Opus 4.5 Now Available`,
  label: 'feature',
  avatar: '/images/samuel_su_profile.png',
  hashtags: ['Claude Opus 4.5', 'AI Models', 'Anthropic', 'Customer Support'],
  thumbnail,
  preview: `We're excited to announce Anthropic's Claude Opus 4.5 on Chat Data. The most advanced Opus model yet, delivering exceptional accuracy and improved efficiency for your customer support.

**Key improvements:**
- Industry-leading accuracy with 80%+ SWE-bench performance
- 48-76% better token efficiency for cost-effective operations
- Superior context understanding for complex conversations

Available now in your Chat Data model selector.`,
};


We're excited to announce the availability of Anthropic's Claude Opus 4.5 on Chat Data. This latest flagship model brings exceptional accuracy, superior reasoning capabilities, and improved efficiency to your customer support chatbots.

## What's New

- **Claude Opus 4.5**: Most advanced Opus model with breakthrough 80%+ SWE-bench accuracy

## Key Improvements for Customer Support

- **Higher Accuracy**: Industry-leading performance with significantly reduced hallucinations
- **Better Cost Efficiency**: 48-76% better token efficiency for high-volume conversations
- **Extended Autonomous Work**: Handles complex, multi-turn customer conversations effortlessly
- **Improved Context Understanding**: Superior conversation flow and information retention

Claude Opus 4.5 is now available in your Chat Data model selector, offering premium performance for businesses that need the highest quality AI responses.

Claude Opus 4.5 Now Available

export const article = {
  author: 'Samuel Su',
  date: '2025-11-19',
  title: `Gemini 3.0 Pro Now Available`,
  label: 'feature',
  avatar: '/images/samuel_su_profile.png',
  hashtags: ['Gemini 3.0 Pro', 'Google DeepMind', 'AI Models', 'Advanced Reasoning', 'Customer Support'],
  thumbnail,
  preview: `Google DeepMind's Gemini 3.0 Pro is now available on Chat Data, featuring state-of-the-art reasoning for complex problem-solving, enhanced customer intent understanding, and optimized performance at the same cost as previous Gemini models. We're currently using the Preview model and will automatically upgrade to the standard version when available.`,
};


We're excited to announce the availability of Google DeepMind's Gemini 3.0 Pro on Chat Data. This flagship model brings state-of-the-art reasoning capabilities and enhanced performance to your customer support chatbots.

## New Model

- **Gemini 3.0 Pro**: Google's most advanced model with exceptional reasoning capabilities and performance across complex tasks

## Key Improvements for Chatbots

- **Advanced Reasoning**: State-of-the-art performance for complex problem-solving and multi-step tasks
- **Enhanced Understanding**: Superior comprehension of customer intent across text and multimodal content
- **Improved Code & Logic**: Better handling of technical support questions and debugging scenarios
- **Optimized Performance**: Faster responses with dynamic thinking levels for different query complexities
- **Same Cost**: Credit cost remains the same as previous Gemini models

## Important Note

We are currently using the **Gemini 3.0 Pro Preview** model. When Google releases the standard Gemini 3.0 Pro model, we will automatically upgrade all chatbots to use the standard version with no action required on your part.

Gemini 3.0 Pro is now available in your Chat Data model selector.

Gemini 3.0 Pro Now Available

export const article = {
  author: 'Yongrui Su',
  date: '2025-11-18',
  title: `Launch: Real-Time PII De-Anonymization for Natural AI Conversations`,
  label: 'feature',
  avatar: '/images/yongrui_su_profile.jpeg',
  hashtags: ['Privacy Protection', 'User Experience', 'PII Anonymization', 'Conversational AI'],
  thumbnail,
  preview: `We've completed the privacy loop with automatic PII de-anonymization! Your AI now delivers natural, readable responses while keeping customer data completely protected from LLM providers.

**Highlights**

🔒 **Full Privacy Protection** - Customer PII never reaches AI models

✨ **Natural User Experience** - Users see clean responses without ugly placeholder tokens

🤖 **Automatic Operation** - Works seamlessly when anonymization is enabled
`,
};


When we launched PII Anonymization, we gave you the power to protect your customers' sensitive data from AI models. But there was a catch: while the AI model never saw real emails, phone numbers, or names, your end users saw responses filled with ugly placeholder tokens like PERSON_7299dd, EMAIL_ADDRESS_217195, and PHONE_NUMBER_862c7c.

Today, we're completing the privacy loop with automatic PII de-anonymization - giving you the best of both worlds with full privacy protection AND natural, readable responses.

---

## The Problem We Solved

When PII Anonymization was enabled, users would see responses filled with ugly placeholder tokens instead of natural language.

**User Query:**
```plaintext
Summary below: Our team members are: Alice, phone 555-1111; Bob, phone 555-2222;
and Carol, phone 555-3333. Their email addresses are: alice@company.com,
bob@company.com, and carol@company.com. Please contact headquarters at phone
number 555-4444.
```

**What Users Saw Before (With Placeholders):**
```plaintext
Here's a cleaned-up, natural-language version of your summary with placeholders
preserved:

Our [PERSON_7299dd] are:

[PERSON_64489c], phone [PHONE_NUMBER_862c7c]
[PERSON_2fc1c0], phone [PHONE_NUMBER_793eec]
[PERSON_150c16], phone [PHONE_NUMBER_517ce1]

Their email addresses are:
[EMAIL_ADDRESS_2353b2]
[EMAIL_ADDRESS_217195]
[EMAIL_ADDRESS_f95f10]

Please contact headquarters at phone number [PHONE_NUMBER_9a5359].
```

The privacy worked perfectly - the AI never saw the real data. But the user experience was terrible.

---

## The Solution

**What Users See Now (Natural Language):**
```plaintext
Here's a cleaned-up, natural-language version of your summary:

Our team members are:

Alice, phone 555-1111
Bob, phone 555-2222
Carol, phone 555-3333

Their email addresses are:
alice@company.com
bob@company.com
carol@company.com

Please contact headquarters at phone number 555-4444.
```

Same privacy protection. Natural user experience. Automatic.

---

## How It Works

The magic happens automatically:

🔒 **What the AI Sees:** PERSON_64489c, EMAIL_ADDRESS_2353b2, PHONE_NUMBER_862c7c

✨ **What Your Users See:** Alice, alice@company.com, 555-1111

Your customer data never reaches the AI provider, but your users always see natural, readable responses.

---

## Why This Matters

🔒 **Full Privacy** - AI never sees real names, emails, or phone numbers

✨ **Natural Conversations** - No more ugly PERSON_7299dd tokens breaking the user experience

🤖 **Zero Setup** - Already works if you have PII anonymization enabled

🚀 **No Delays** - Happens instantly during streaming

Whether you're in e-commerce handling order details, healthcare protecting patient information, financial services managing sensitive accounts, or customer support - your users now get clean, readable responses while their data stays completely protected.

---

## Getting Started

De-anonymization works automatically with nothing to configure!

If you already have Anonymize PIIs for LLM enabled in Settings then Privacy, de-anonymization is now active. Your users will see natural responses instead of placeholder tokens while your customer data stays fully protected from AI models.

If you haven't enabled PII anonymization yet, head to Settings then Privacy and toggle on Anonymize PIIs for LLM to start protecting customer data with automatic de-anonymization built in.

Launch: Real-Time PII De-Anonymization for Natural AI Conversations

export const article = {
  author: 'Yongrui Su',
  date: '2025-11-16',
  title: `AI Search Bar & Enhanced Widget Experience`,
  label: 'improvement',
  avatar: '/images/yongrui_su_profile.jpeg',
  hashtags: ['Widget', 'UX', 'Page Context', 'AI Search'],
  thumbnail: searchBarWidget,
  preview: `
Transform your website's AI experience with the new AI Search Bar widget and powerful contextual awareness features that make customer interactions more intuitive and intelligent.

**Highlights**

🔍 **AI Search Bar Widget** - New sleek search bar display style alongside the traditional chat bubble for a seamless on-page experience

⚡ **Direct Interaction** - Conversation starters now appear directly on your website—no need to open the widget first

📝 **Ask AI About Selection** - Customers can select any text on your page and instantly ask AI about it with one click

🧠 **Page Awareness by Default** - Chatbot automatically understands page content, navigation, and user context for personalized responses

🔒 **Privacy Protected** - Built-in PII anonymization keeps personal information safe while maintaining conversation context

📱 **Full-Screen Mode** - Users can expand the widget to full-screen size for a more immersive chat experience on desktop
`,
};


We're excited to launch major improvements to the Chat Data widget experience—making it easier for your customers to interact with your AI chatbot and get personalized, context-aware responses instantly.


---

## AI Search Bar Widget Style

You now have **two display styles** to choose from when embedding your chatbot on your website:

🔵 **Chat Bubble** - The familiar floating chat bubble that appears in the corner of your page, perfect for persistent support access.

🔍 **AI Search Bar** - A sleek, modern search bar that blends naturally into your page layout, ideal for knowledge bases, documentation sites, and content-heavy pages where users expect to search for information.

Both styles provide the same powerful AI capabilities—choose the one that best matches your website's design and user expectations. You can configure this in **Settings > Chat Interface > Display Style**.

---

## Direct Conversation Starters

Say goodbye to the extra click! Conversation starters (suggestion questions) now appear **directly on your website** instead of only inside the chatbot interface.


⚡ **Instant Engagement** - Customers can click a suggested question right from your page to start a conversation without opening the widget first.

🎯 **Lower Friction** - Reduce the barrier to engagement by showing relevant questions upfront based on the current page context.

💡 **Better Discovery** - Visitors immediately see what your chatbot can help with, increasing interaction rates and improving the customer experience.

Previously, users had to open the chatbot widget and then click on conversation starters inside the chat interface. Now they can engage instantly with one click.

---

## Enhanced Suggestion Messages Control

We've separated the controls for suggestion message generation and visibility, giving you more flexibility to optimize both AI costs and user experience across different devices.

**Three Independent Controls:**

🤖 **Dynamic Suggestion Generation** - Control how many AI-generated suggestion questions to create based on conversation context (1-5 suggestions). This determines how many questions the AI will generate, which impacts your AI usage costs.

💻 **Visible Suggestions (Desktop)** - Control how many suggestion messages display when the widget is closed on desktop screens. This is purely a display setting and doesn't affect AI generation.

📱 **Visible Suggestions (Mobile)** - New mobile-specific setting for narrow screens. Choose a smaller number of suggestions to display on mobile devices where screen space is limited, while still generating the full set in the background.



**Why This Matters:**

- **Cost Optimization** - Generate fewer suggestions to reduce AI costs without changing what users see
- **Mobile Experience** - Show fewer suggestions on mobile to avoid cluttering small screens
- **Flexibility** - Configure generation and visibility independently based on your needs

**Example Use Case:** You might generate 5 dynamic suggestions for better variety, but only show 3 on desktop and 2 on mobile to keep the interface clean.

Configure these settings in **Settings > Chat Interface > Suggestion Messages**.

---

## Ask AI About Selected Text

Give your customers the power to ask about **any content on your page** with the new "Ask AI" feature.


📝 **Text Selection Magic** - When a user selects any text on your website, an "Ask AI" button appears automatically above the selection.

💬 **Instant Context** - Clicking the button opens the chatbot with the selected text pre-populated, ready for questions like "Explain this," "What does this mean?" or "Tell me more about this."

🎓 **Perfect for Education & Documentation** - Ideal for technical docs, educational content, product descriptions, or any site where users might need clarification on specific passages.

This feature works seamlessly with both the chat bubble and AI search bar widget styles.

---

## Page Awareness by Default

Your chatbot is now **automatically aware of the page context** when integrated through the widget—no configuration needed.

🧠 **Smart Context Understanding** - The chatbot receives key information about the current page including:
- Navigation links and site structure
- Important factual information visible on the page
- User information (if available)
- Current page URL and context

🎯 **Personalized Responses** - By combining your knowledge base with real-time page content, the chatbot delivers more accurate, contextually relevant answers tailored to what the customer is currently viewing.

💡 **Better Customer Experience** - Instead of relying solely on static knowledge base content, your chatbot understands where the customer is on your site and what they're looking at.

### Privacy Protection

Concerned about sensitive information from page content reaching AI models? We've got you covered:

🔒 **Automatic PII Anonymization** - Enable "Anonymize PIIs for LLM" in **Settings > Privacy** to automatically mask personal information (emails, phone numbers, credit cards, etc.) from page content before it's sent to AI models.

⚙️ **Full Control** - Want to disable page context entirely? Simply turn off the "Enable Page Context" option in **Settings > Chat Interface > Other Configuration** to stop tracking page content and URLs.

**Note:** Page awareness is enabled by default for all chatbots using the widget integration. You don't need to do anything to turn it on—it just works.

---

## Full-Screen Widget Mode

Give your customers more space to interact with your AI chatbot with the new full-screen expansion option.


📱 **Desktop Optimization** - Users can expand the chatbot widget to fill their entire screen for a more immersive chat experience, perfect for complex conversations or when reviewing detailed responses.

🔄 **Easy Toggle** - One click expands to full-screen, another click returns to the original widget size—simple and intuitive.

💼 **Professional Experience** - Great for enterprise users, support conversations, or when customers need to reference extensive documentation or product details in the chat.

This feature is available for both chat bubble and AI search bar widget styles on desktop devices.

---

## Getting Started

All these features are **available now** for your Chat Data chatbots:

1. **Choose Your Widget Style** - Go to **Settings > Chat Interface > Display Style** to select between chat bubble and AI search bar
2. **Configure Page Context** - Visit **Settings > Chat Interface > Other Configuration** to manage page awareness settings
3. **Enable Privacy Protection** - Head to **Settings > Privacy** to turn on PII anonymization for page content
4. **Test the Experience** - Use the embedding simulator to see how conversation starters, text selection, and full-screen mode work together

These enhancements work seamlessly across all your existing chatbot configurations and integrations. No migration or code changes required—just update your settings and start delivering a better customer experience today.

Changelog

Create Chatbots with your data

In just a few minutes, you can craft a customized AI representative tailored to yourself or your company.