AI Chatbot Hub
HomepageAPI docsDiscordLog inSign up
Documentation
Documentation
  • welcome
    • Quick start guide
    • Creating your first AI chatbot
  • Dashboard
    • Analytics
  • Account
    • Account settings
    • Get OpenAI API key
    • API docs
  • Inbox
    • Navigating the Inbox menu
  • Sources
    • Navigating the Sources menu
    • Best practices for preparing data
    • Adding training data
    • How to work with tables
  • AI agents
    • Creating AI agents
    • Fine-tune intents
    • Supervisor overrides
    • Function calling
    • Bias and stickiness
  • Customizations
    • Settings
      • Security
      • Conversations
      • Custom domains
      • Disclaimer
      • Human support escalation
    • User Data Management
      • File uploader
      • Conversation labeling
      • User data collection
      • Forms
      • Webhooks
      • User identity verification
    • Appearance & Style
      • General
      • Messages
      • Suggested prompts
      • Watermark
      • Button
  • Collaborators
    • Permissions
    • Transfer ownership
  • Preview
    • Overview
  • Go live
    • Integrations
  • Community
    • Partners
    • Affiliates
    • Discord
  • Resources
    • Get a free template
    • Enterprise
    • Glossary
    • Support
Powered by GitBook
On this page
  • Set up and configuration
  • Testing
  • Browser notification permissions
  1. Customizations
  2. Settings

Human support escalation

Activate human support in your AI chatbot with AI Chatbot Hub's Human Escalation feature, enabling real-time assistance and boosting user satisfaction.

PreviousDisclaimerNextUser Data Management

Last updated 4 months ago

While AI chatbots are highly effective, sometimes human expertise is still essential. AI Chatbot Hub’s human support escalation feature lets you or your team step into any active chat session, taking over the conversation when needed.

Set up and configuration

In our multi-agent system, the human support agent functions like other agents, but instead of generating responses with an LLM, it routes the user's request to a human. To enable this feature, go to Customizations > Settings. Toggle on the feature.

There are a few functionalities to be configured for the Human support escalation feature:

  • Human agent requested message: the message that will be sent to the user when the AI supervisor detects that a human agent's help is needed

  • Notification sound: enable to hear a notification sound when a new support chat request is received

  • Live chat starts announcement: the message the user will see when a new support agent has taken over the conversation.

  • Live chat ends announcement

  • AI output alert triggers: add up to 10 messages for the AI to monitor

Don't forget to upload a profile picture to your account so your users know who they are interacting with.

Testing

To test your human support escalation, generate a public link to the chatbot and open it in another browser tab.

Note that the “Preview” menu doesn’t support human escalation testing. Go to the “Go Live” menu and click “Link".

With the link, open a new browser tab to access the chatbot as a user and test it by sending requests for human support.

To respond as a human agent, go to the “Inbox” in your AI Chatbot Hub tab, where an orange dot will mark sessions needing human support. Click on the "Reply as a human" text box at the bottom.

Once you send your first message, AI Supervisor and all AI Agents will pause, allowing you to converse directly with the user.

When you’re finished, click “End Live Chat” to return the user to the AI chat and end the human support. Now the user will continue chatting with the AI, while the conversation is marked as resolved.

Adjust the criteria for human redirection as needed by fine-tuning the configurations to your taste.

Browser notification permissions

Each browser requires permissions for notifications and sounds by website. Depending on your browser, make sure you enable both in your browser settings to receive notifications smoothly.

Human support escalation has been received
View from the human support agent replying from "Inbox"
View from the user's perspective
enable human escalation
human support escalation settings
human support escalation notification
human support escalation
chat view from user's perspective