Imagine you want your AI Assistant to start conversations by addressing each user by their first name and crafting the welcome message in their particular language. Hardcoded or static messages make this impossible.
We've introduced the "Start" flow to make this possible. This flow is a "system" flow. It comes with each assistant and can't be deleted, but it can be se to offline or to draft mode, in which case the AI Assistant's welcome message will automatically fallback to the configuration set in the "Chat Customization" settings. Also, if your Start flow is live but misconfigured (e.g.: nodes not connected), it will also fallback.
Configuring the Start Trigger Execution type
The start trigger can be configured in two different modes for live chat:
Widget loaded: The start flow will be executed once the widget is loaded. The user won't need to open the widget for the AI Assistant to execute the flow.
Widget opened: The start flow will be executed only when the user opens the widget.
The Start Trigger in Action
In this scenario we used the users' first_name (leveraging identity verification) and their browsers language, which is automatically detected using Frontline live chat widget metadata capture, to craft a personalized welcome message for each user.
After the first message is sent by the AI assistant in this particular flow, the flow is "exited" because there aren't any other nodes to execute. But users can craft highly sophisticated flows, for example you may want your AI Assistant to:
Start the conversation by requesting lead information to qualify the lead before addressing other questions.
Request a candidates contact information before starting an AI driven interview process.
The possibilities are endless.