WhatsApp conversations are recorded alongside all your other chat interactions. From your dashboard you can review full message history, inspect media attachments, monitor AI performance, and track costs per conversation.
Viewing WhatsApp conversations
WhatsApp conversations appear in the Conversations section with a green “WhatsApp” badge. Each record includes:
- Complete message history between customer and AI
- Customer phone number
- Connected AI assistant
- Timestamps for every message
- Cost and token usage metrics
How a conversation unfolds
- A customer sends a message to your WhatsApp number.
- The system creates or retrieves the existing conversation for that number.
- The message is routed to your AI assistant.
- The AI generates and sends a reply.
- The 24-hour messaging window opens.
Within the 24-hour window
- The AI responds automatically to each incoming message.
- Free-form replies are permitted — no template required.
- The conversation continues naturally.
After 24 hours
- If the customer hasn’t messaged, the window closes.
- To re-engage, send an approved template message.
- The AI remains ready to respond as soon as the customer replies.
Image analysis (Vision)
When a customer sends an image, your AI assistant can analyze it using vision-capable models (OpenAI, Claude, Gemini). The AI’s interpretation and response appear in the conversation thread alongside the original image.
Voice note transcription
Incoming voice messages are automatically transcribed using your assistant’s configured language. The transcribed text appears in the conversation alongside the audio, and the AI responds to the transcribed content.
Attachments gallery
All inbound media — images, audio, video, and documents — are stored as attachments and viewable from the conversation detail view.
Sending files to customers
Operators can send files directly to customers from the conversation view, which is useful for sharing documents or resources during an active conversation.
AI assistant behavior
System prompt
Your assistant’s system prompt applies across all channels — voice, web widget, and WhatsApp. If your assistant handles both voice and text, consider adding instructions for text-friendly formatting:
Include guidance in your system prompt for WhatsApp conversations — for example, instructing the AI to use short paragraphs and bullet points instead of long prose.
- Looking up information in databases
- Scheduling appointments
- Transferring to a human agent
- Collecting and validating customer data
Knowledge bases
Your assistant’s knowledge bases are available in WhatsApp conversations, allowing the AI to answer questions from your uploaded content.
Managing conversations
Filtering
Filter the conversation list by:
- Type — WhatsApp, Web Widget, or Test
- Assistant — View conversations handled by a specific assistant
- Date range — Narrow to a specific period
- Customer identifier — Find conversations by phone number
Exporting data
- Navigate to Conversations.
- Apply any filters you need.
- Use the export option to download conversation data.
Conversation metrics
| Metric | Description |
|---|
| Total conversations | Number of unique customer conversation threads |
| Messages per conversation | Average message count per thread |
| Response rate | Percentage of customer messages that received an AI response |
| Average response time | Time between customer message and AI reply |
| Cost per conversation | Average total cost including AI model usage |
Best practices
- Review regularly — Check AI responses to catch inaccurate information, wrong tone, or failed interactions early.
- Improve your assistant — Use common questions from conversations to update knowledge bases, refine the system prompt, and add missing tools.
- Monitor response latency — WhatsApp users expect fast replies. Watch for delays or timeouts.
- Plan for escalations — Configure your assistant to transfer to a human agent, collect a callback number, or schedule a follow-up call when a conversation requires it.
Next steps