Live Chat: Perfect Collaboration between AI and Humans
DMflow.chat’s Live Chat system aims to seamlessly integrate automation and manual services. When the bot cannot solve a problem, or triggers specific business needs, we can smoothly transfer the conversation to a human agent.
Two Customer Service Modes
You can choose the operation mode according to your needs:
- DMflow Omni-channel Customer Service (Recommended): Unified reply to messages from all channels like LINE, Messenger, Telegram, etc., on the DMflow platform. The advantages are a unified interface, centralized data, and the ability to use AI-assisted functions.
- Direct to Official Platform: Redirect conversations back to the official native backend of Messenger or LINE.
- Pros and Cons Comparison: Native backends have the most complete functions (such as LINE’s phone function, sending special files), but lack the convenience of API integration and cross-platform management. You can choose to mix them depending on the situation.
DMflow Customer Service Platform Highlights
Inbox Management
- Assignment Mechanism: Supports “assigning” conversations to specific agents, or leaving them in the “Unassigned” area waiting to be claimed.
- Internal Note: Agents can leave internal notes next to the conversation, which visitors will not see. Supports starting with
/noteand using@to call other users to view.
Contacts and Permissions
- Dedicated Inbox: Agents and tenant creators have dedicated inboxes and can directly contact DMflow official customer service.
- User Attribute Editing: Agents can directly modify user attributes and tags in the bot’s memory.
- Note: Tags here refer to “Bot Memory Tags”, currently the ticket tagging system specific to Agents is not yet implemented.
Agent Skills
Agents not only reply to messages but can also directly call “Domain Skills”:
- Doc/Chat Skills: Directly use AI’s knowledge base to assist in answering.
- Form Skills: Clicking a form opens a new window and automatically substitutes “Visitor Mode” parameters. This means agents can view form data created by that visitor and even assist in filling it out.
Customer Service Interface Operation Guide

- Search and Notification: Supports searching visitors. After enabling audio notifications, a prompt sound will be played when a new visitor calls.
- Filter: Can filter by “Assigned to me”, “Unassigned”, “All”.
- Conversation Control:
- Extend Service Time: Avoid Session expiration.
- Complete/Close: “Closing” the conversation formally ends this customer service activity.
- Load Conversation: Scrolling itself automatically loads, this button just helps scroll up to automatically load historical messages.
- Assign: Transfer conversation to other agents.
- AI Translation Magic Wand: Click to automatically translate visitor messages into your browser’s language.
- AI Suggestion (Copilot): Displays AI-suggested reply content and steps.
- Input Box Toolbar (Left to Right):
- Emoji: Emojis.
- Shared Note: For internal communication.
- Chat Magic Wand: Translate your reply into the other party’s language and send.
- Default Template: Quick reply canned messages.
- Warning Reminder: Displays channel limits (e.g., LINE Push requires payment, Messenger/IG 24-hour reply limit, etc.).
- Sidebar Info: Switch to view user attributes, skill panel, notes, or AI suggestions.
- Block Status: Displays whether the visitor is blocked by you (not blocked by the other party). Web chat can display the other party’s online status.
- User Tags: View and manage visitor tags.
Special Considerations for LINE Channel
When the conversation is transferred to a human agent, the visitor’s message will carry a reply token.
- Free Quota: If you reply within a short time, the system will prioritize consuming the cached
reply token, at which time it does not count towards Push fees. - Paid Push: If you answer multiple messages consecutively, or exceed the token validity period, subsequent messages will automatically switch to Push mode, which may incur additional fees.
Supplement: Although LINE and Telegram group customer service is supported, using LINE groups is not recommended because every message in the group will consume your message quota in multiples of the group size.