Chatbot Nodes

Chatbot nodes let you build conversational chatbots that can answer questions, route conversations, remember context, and hand off to human agents. There are 7 chatbot nodes in total.

Chatbot nodes work together as a system. A typical chatbot workflow starts with a Chat Trigger, routes through a Chat Router, uses Chat Memory to maintain context, processes data through a Data Scope for security, formats output with Message Format, sends responses via Chat Response, and escalates when needed using Human Handoff.

Chat Trigger

Icon: MessageCircle | Category: Logic

The entry point for chatbot workflows. Fires when a user sends a message through any connected channel (web widget, SMS, etc.).

Inputs: None (this is a starting node)

Outputs: data-out (message text, sender info, channel, timestamp)

Config:

Channels: Which channels to listen on
Greeting: Optional auto-reply when a conversation starts
Debounce: Wait time before processing (to batch rapid messages)

Example: When a customer sends a message on the website chat widget, fire the chatbot workflow.

Chat Response

Icon: MessageSquareReply | Category: Template

Send a formatted response back to the chat user. Supports plain text, rich cards with images, quick-reply buttons, and carousels of multiple cards.

Inputs: data-in (response content)

Outputs: data-out (confirmation of sent message)

Config:

Type: Text, rich card, quick replies, or carousel
Content: The response text or card content
Quick replies: Optional button labels for suggested responses
Actions: Optional clickable action buttons

Example: Send a rich card showing the customer's next booking with date, time, and "Reschedule" and "Cancel" quick-reply buttons.

Chat Router

Icon: Route | Category: Logic

Route conversations to different paths based on what the user is asking about. Supports keyword matching, regular expressions, and AI-powered intent classification.

Inputs: data-in (chat message)

Outputs: One output per route, plus a fallback for unmatched messages

Config:

Routes: List of intent routes with matching criteria
Match type: Keyword, regex, or AI classification
Fallback: Where unmatched messages go
Confidence threshold: How confident the AI must be (for AI classification)

Example: Route "What's my next appointment?" to the booking lookup flow, "I need to cancel" to the cancellation flow, and everything else to the general AI assistant.

Chat Memory

Icon: Brain | Category: Data Ops

Store and retrieve conversation context across multiple messages. Keeps track of what has been discussed so the chatbot can reference earlier parts of the conversation.

Inputs: data-in (conversation data)

Outputs: data-out (data enriched with conversation history)

Config:

Operation: Get (retrieve memory), set (store new data), append (add to existing), or clear (reset)
Key: What to store/retrieve (e.g., "customerid", "bookingtopic")
TTL: How long to keep the memory before it expires (in minutes)

Example: When a customer mentions their booking number, store it in memory. Later messages can reference it without asking again.

Data Scope

Icon: Shield | Category: Logic

A security gate that controls what database data the chatbot can share. Prevents the bot from accidentally exposing sensitive customer information.

Inputs: data-in (full data from database queries)

Outputs: data-out (filtered, safe data)

Config:

Mode: Allowlist (only pass listed fields) or denylist (block listed fields)
Fields: Which fields to allow or block
PII detection: Automatically detect and filter common PII fields
Role-based: Different filtering rules based on the user's role

Example: Before the chatbot shares booking details, use Data Scope to strip out payment card numbers and internal notes.

Message Format

Icon: LayoutList | Category: Template

Format raw data into chat-friendly message structures. Turns database records and API responses into readable messages.

Inputs: data-in (raw data to format)

Outputs: data-out (formatted message ready for Chat Response)

Config:

Format: Text, markdown, rich card, list, table, or receipt
Template: How to arrange the data fields in the output
Truncate: Maximum length before truncating

Example: Format a booking record into a readable message: "Your next appointment is on March 15 at 2:00 PM for Deep Cleaning at 123 Main Street."

Human Handoff

Icon: Headphones | Category: Logic

Escalate the conversation from the chatbot to a live human agent. The bot stops responding and a human takes over.

Inputs: data-in (conversation context and history)

Outputs: data-out (handoff confirmation)

Config:

Department: Which team should handle the conversation (support, billing, technical)
Priority: Urgency level (low, normal, high, urgent)
Context message: Summary of the conversation for the human agent
Queue message: What to tell the customer while they wait

Example: When the chatbot detects the customer is frustrated or the issue is too complex, hand off to a live support agent with a summary of the conversation so far.