Action Map

The Action Map is a powerful visual workflow designer that lets you create custom conversation flows for your Support BV. Design exactly how your assistant responds to different scenarios, guide users through multi-step processes, and create sophisticated automation—all without writing code.

What is the Action Map?

The Action Map is a node-based workflow designer that gives you complete control over your Support BV's conversation logic. Each node represents a specific action (like showing a message, checking a condition, or gathering knowledge), and you connect these nodes to create the exact conversation flow you want.

Key Capabilities:

Visual Design: Drag, drop, and connect nodes on an intuitive canvas
Real-Time Testing: Preview changes instantly with the built-in test chatbot
Version Control: Save drafts, publish changes, and rollback if needed
Template System: Save and reuse workflows across different Support BVs
Multi-Channel: Workflows work across website, messaging platforms, and more

Quick Setup

Access the Action Map

Navigate to Support BV > Action Map from your dashboard.

Explore the Default Workflow

Review the pre-built workflow that handles welcome messages and questions. This gives you a starting point to understand how workflows operate.

Customize Your Flow

Add, remove, or modify nodes to match your specific needs. Connect nodes by dragging from output handles to input handles.

Test & Publish

Use the test chatbot (bottom right) to preview your changes, then click Publish to make them live.

Understanding the Default Workflow

Every new Support BV starts with a smart default workflow that handles common scenarios:

Start Node

The entry point for all conversations. The Start Node receives user messages and prepares them for processing.

Check Flow Type (Condition)

A Condition node that routes conversations based on whether the user is:

Opening the chat for the first time (session_opened_first_message)
Returning to an existing session (session_opened_other_message)
Asking a normal question (everything else)

Branch 1: First-Time Welcome

For first-time visitors:

Answer Node displays a welcome message
Buttons Node shows helpful quick-action buttons

Branch 2: Welcome Back

For returning users:

Answer Node displays a welcome back message

Branch 3: Question Handling

For normal questions:

Knowledge Node retrieves relevant information from training data
Answer Generator Node creates an AI response using the retrieved knowledge
Answer Node displays the generated response to the user

Customizable Default

The default workflow is fully customizable. You can modify messages, add steps, or completely redesign the flow to match your needs.

Available Nodes

The Action Map provides specialized nodes for building sophisticated conversations. Nodes are organized into categories:

These custom nodes are unique to BrandVox and power your Support BV's intelligence:

Answer Node

Purpose: Display messages to users

The only node that can show text to users. Use Answer nodes to:

Welcome users
Provide information
Confirm actions
Display AI-generated responses

Configuration:

Message text (supports variables like {{question}} or {{nodeId}})
Variable substitution from previous nodes

Example Use:

Welcome back! How can I help you today?

Answer Generator Node

Purpose: Create AI-powered responses

Generates intelligent answers using your:

Personality settings
Brand voice
AI model and temperature
Context from previous nodes

Configuration:

Question input (typically {{question}})
Context input (typically from Knowledge node)
Uses workspace AI settings automatically

Example Use: Connect a Knowledge node to provide context, then use Answer Generator to create a personalized response based on that knowledge.

Knowledge Node

Purpose: Retrieve relevant information

Searches your assigned training materials to find information relevant to the user's question.

Configuration:

Question input (typically {{question}})
Automatically searches assigned training content

How it Works:

Receives the user's question
Searches through all training materials assigned to this Support BV
Returns the most relevant information
Passes results to Answer Generator or other nodes

Condition Agent Node

Purpose: AI-powered decision making

Uses AI to analyze questions and route conversations based on intent, sentiment, or content.

Configuration:

Instructions for the AI
Output branches for different scenarios
Question input

Example Use: Route technical questions to detailed troubleshooting, sales questions to product info, and complaints to a specialized support flow.

Buttons Node

Purpose: Display interactive buttons

Shows clickable buttons in the chat interface for quick user actions.

Configuration:

Button text
Button style (primary, secondary, subtle)
Multiple buttons per node

Limitations:

Currently displays in chat bubble and iframe only
Not available in third-party integrations (Messenger, Instagram, etc.)

Example Use:

Button 1: "How can I contact support?" (primary)
Button 2: "What are your hours?" (secondary)
Button 3: "Track my order" (subtle)

Leads Node

Purpose: Capture user information

Displays a form to collect contact details from users.

Configuration:

Full name field (optional/required)
Email field (optional/required)
Phone field (optional/required)

Limitations:

Currently displays in chat bubble and iframe only
Not available in third-party integrations (Messenger, Instagram, etc.)

Data Collection: All submitted leads are available in Support BV → Leads for viewing and export.

Human Support Node

Purpose: Detect and escalate to human agents

Automatically detects when users want to speak with a real person and escalates the session to human support. Uses AI-powered detection to identify requests like "talk to a human", "speak to an agent", or "I need a real person".

Configuration:

Enable Human Support: Toggle to enable/disable the feature
Detection Sensitivity: Choose how sensitive the AI should be:
- High: Detects most variations and subtle hints
- Medium: Balanced detection (recommended)
- Low: Only clear, explicit requests
Human Support Message: Custom message shown when escalation occurs
Enable Lead Form: Optionally collect user information before escalating (web/websocket only)

Lead Form Fields (Optional):

Full Name (enable/require/customize label and placeholder)
Email (enable/require/customize label and placeholder)
Phone Number (enable/require/customize label and placeholder)
Form Title and Submit Button Text

How It Works:

User sends a message requesting human support
AI detects the intent based on configured sensitivity
For Web/Websocket with Lead Form enabled:
- Shows lead capture form to collect user information
- After submission (or if no required fields), escalates to human agent
For Non-Web Platforms or Lead Form disabled:
- Immediately shows escalation message
- Notifies human agents in Conversations page
Session is marked as "Human Support" type
Human agents can respond via the Conversations page

Platform Behavior:

Web/Websocket: Can show lead form before escalation
Messenger/Instagram/WhatsApp/Telegram: Immediate escalation (no lead form)

Detection Examples: The AI intelligently identifies variations like:

"I want to speak to a human"
"Can I talk to a real person?"
"Connect me to an agent"
"Is there someone I can talk to?"
"Transfer me to support"

Best Practices:

Start with Medium sensitivity and adjust based on results
Customize the escalation message to match your brand voice
Use lead forms on web to collect context before escalation
Monitor the Conversations page regularly for escalated sessions
Configure your team's agent display names for consistency

Managing Human Support Sessions: After escalation, human agents can:

View escalated sessions in Support BV → Conversations (filter by "Human" type)
Send real-time markdown-formatted responses
Switch sessions back to AI if appropriate
View user information captured via lead forms

Essential nodes for controlling conversation flow:

Start Node

The entry point for all conversations. Required in every workflow.

Condition Node

Routes conversations based on conditions:

String comparisons (equals, contains, starts with, etc.)
Number comparisons (greater than, less than, etc.)
Boolean logic
Multiple conditions with "and"/"or" logic

Loop Node

Repeat actions for multiple items or iterations.

Advanced nodes for specialized use cases:

LLM Node

Purpose: Send prompts to AI language models and get responses

Use the LLM node to directly communicate with AI models for custom analysis, generation, or processing tasks.

Configuration:

Model Selection: Choose from available AI models (GPT, Claude, Gemini, Grok, etc.)
Messages: Configure conversation context with System, User, Assistant, or Developer roles
Memory Management: Control conversation history with options:
- All Messages: Include full conversation history
- Window Size: Use last N messages only
- Conversation Summary: Summarize entire conversation
- Conversation Summary Buffer: Summarize when token limit reached
Structured Output: Define JSON schema for predictable response format
Flow State Updates: Store LLM responses in workflow variables

Example Use Cases:

Custom content analysis (sentiment, intent, categorization)
Data extraction from user input
Multi-step reasoning tasks
Custom response generation with specific formatting

How It Works:

Accepts user input and configured messages
Sends to selected AI model
Returns AI-generated response
Can update workflow state with the response
Outputs content to next node

Best Practice: Use LLM nodes when you need custom AI behavior beyond what Answer Generator provides, such as structured data extraction or complex multi-step analysis.

Agent Node

Purpose: Create AI agents that can use tools and make decisions

The Agent node enables sophisticated AI workflows by allowing the AI to dynamically choose and use tools during execution, enabling multi-step reasoning and autonomous problem-solving.

Configuration:

Model Selection: Choose AI model (must support function calling)
Tools: Select external tools the agent can use:
- Search tools
- Calculators
- API integrations
- Custom tools
Knowledge Bases: Connect to document stores or vector databases for retrieval
Built-in Tools: Platform-specific capabilities:
- OpenAI: Web Search, Code Interpreter, Image Generation
- Gemini: URL Context, Google Search
- Anthropic: Web Search, Web Fetch
Memory & State: Same conversation memory options as LLM node
Instructions: Define the agent's behavior and decision-making guidelines

Example Use Cases:

Research assistant that can search and synthesize information
Technical support agent with access to knowledge base and troubleshooting tools
Data analysis agent with calculator and chart generation
Multi-step problem solving requiring tool use

How It Works:

Receives user input and configured instructions
Analyzes what tools/information are needed
Calls relevant tools autonomously
Processes tool results
May call additional tools based on results (recursive tool calling)
Returns final synthesized response

Key Features:

Dynamic Tool Selection: AI chooses which tools to use based on context
Multi-Step Reasoning: Can chain multiple tool calls together
Knowledge Integration: Access document stores and vector databases
Recursive Processing: Continues working until task is complete

Best Practice: Use Agent nodes when tasks require external information, calculations, or multiple steps. The AI will autonomously determine the best approach to solve the user's request.

HTTP Request Node

Make external API calls during conversations.

Code Node

Execute custom JavaScript for complex logic.

Building Your Workflow

Adding Nodes

Click the + Add Node button (top left of canvas) to open the node library. Nodes are organized by category for easy browsing.

Steps:

Click + Add Node
Browse or search for the node type you need
Click the node to add it to the canvas
Position it where you want

Connecting Nodes

Create conversation flows by connecting node outputs to inputs:

How to Connect:

Hover over a node's output handle (right side)
Click and drag from the output handle
Release over the target node's input handle (left side)
The connection is created

Visual Feedback:

Valid connections highlight in green
Invalid connections show red or won't connect
Conditional outputs are numbered (0, 1, 2, "Else")

Connection Rules

Some nodes have specific connection requirements. For example, Answer Generator requires a context input from Knowledge or other data sources.

Editing Node Properties

Customize each node's behavior by editing its properties:

Steps:

Double-click any node
Edit the configuration panel that appears
Enter values, select options, or write messages
Changes save automatically
Click outside or press ESC to close

Common Edits:

Answer Node: Update message text
Condition Node: Modify comparison logic
Buttons Node: Change button text and styles
Leads Node: Toggle required fields

Canvas Controls

Action	How To
Zoom In/Out	Mouse wheel or zoom controls (bottom right)
Pan Canvas	Click and drag empty space
Select Node	Click the node
Move Node	Drag the node
Delete Node	Select and press Delete or Backspace
Delete Connection	Click the connection line and press Delete
Snap to Grid	Toggle grid snap icon (bottom controls)

Keyboard Shortcuts

Delete/Backspace: Remove selected nodes or connections
Ctrl/Cmd + Z: Undo
Ctrl/Cmd + Y: Redo
Mouse Wheel: Zoom in/out

Template Management

Save time and maintain consistency by using templates across your Support BVs.

Browse Pre-Built Templates

Access professionally designed workflow templates for common use cases:

How to Use:

Click the purple template button (top right)
Navigate to Marketplace tab
Browse available templates
Click Preview to see the workflow
Click Import to add it to your Action Map

What Happens:

Template nodes are imported with new IDs
All connections are preserved
Your current workflow remains until you save
You can modify the imported template immediately

Access Your Saved Templates

View and reuse workflows you've saved:

Features:

All your saved templates in one place
Search and filter templates
Preview before importing
One-click import to current Action Map

How to Use:

Click the purple template button (top right)
Navigate to My Templates tab
Find the template you want
Click Import to use it

Save Your Workflow as Template

Preserve your current workflow for reuse:

Steps:

Click the purple template button (top right)
Navigate to Save tab
Enter a template name
Add a description (optional but recommended)
Click Save Template

Best Practices:

Use descriptive names ("E-commerce FAQ Flow", "Lead Capture + Follow-up")
Document what the template does in the description
Save templates at milestone points
Version your templates if making major changes

Upload JSON Templates

Import workflows from JSON files:

Steps:

Click the purple template button (top right)
Navigate to Import tab
Click Upload File or drag & drop
Select your .json file
Click Import

Use Cases:

Share workflows between team members
Migrate workflows between workspaces
Backup complex workflows
Import community templates

Compatibility

Only import templates exported from BrandVox Action Map. Templates from other sources may not work correctly.

Download Workflow as JSON

Export your current workflow for backup or sharing:

Steps:

Click the purple template button (top right)
Navigate to Export tab
Review the JSON preview
Click Download to save as file
Or click Copy to copy to clipboard

Exported Data Includes:

All nodes and their configurations
All connections between nodes
Workflow structure and layout
Node positions on canvas

Credentials Removed: All sensitive data and credentials are automatically stripped from exports for security.

Version Management

The Action Map includes powerful version control to help you work safely and confidently.

Save Draft

Save your work without affecting the live chatbot:

When to Use:

Working on significant changes
Not ready to go live yet
Want to test thoroughly first
Multiple editing sessions needed

How it Works:

Make changes to your workflow
Click Save (top right)
Changes are saved as a draft
Live chatbot continues using the published version
You can continue editing the draft

Visual Indicator: The Save button shows "Unsaved Changes" when you have unpublished modifications.

Draft Auto-Save

Drafts are automatically saved periodically to prevent data loss. Manual saves ensure your latest changes are captured immediately.

Test Your Changes

Preview workflow changes before publishing:

Bottom Right Chatbot Bubble

The floating chatbot bubble at the bottom right tests your draft version:

Features:

Uses your draft workflow immediately after saving
Full conversation simulation
Real Knowledge and Answer Generator integration
Button and Leads form testing
Multiple conversation scenarios

How to Use:

Save your draft changes
Click the chatbot bubble (bottom right)
Start a conversation
Test different paths through your workflow
Verify messages, buttons, and responses

Perfect For:

Quick functionality checks
User experience testing
Conversation flow verification

Publish Changes

Make your draft workflow live for all users:

Steps:

Test thoroughly using both test methods
Click Publish (top right, next to Save)
Confirm the publish action
Wait for confirmation (typically 2-3 seconds)

What Happens:

Draft becomes the new published version
All users immediately get the updated workflow
Previous published version is saved to version history
Draft and published versions are now identical

Rollout Time: Changes are typically live within 2-3 seconds across all channels.

Publish Carefully

Publishing immediately affects all users across all integration points (website, messaging platforms, etc.). Always test thoroughly first.

Version History & Rollback

Access previous published versions and restore if needed:

Viewing History:

Click Version History (top menu)
See all previously published versions
View timestamp and author for each version
Preview workflows (read-only)

Rolling Back:

Find the version you want to restore
Click Rollback on that version
Confirm the rollback action
The selected version becomes published immediately
A new version history entry is created

What Gets Rolled Back:

Entire workflow structure
All node configurations
All connections
Node positions on canvas

Important Notes:

Rollback creates a new published version (doesn't delete history)
Your current draft is preserved
Rollback is immediate—affects live users instantly
You can rollback to any previous published version

Safe Experimentation

The version control system lets you experiment confidently. If something goes wrong after publishing, you can instantly rollback to a known-good version.

First-time visitor flow
Returning user flow
Common question scenarios

Test Edge Cases

Check less common but important scenarios:

Empty or very short questions
Very long questions
Special characters or emojis
Multiple questions in one message

Test Conditional Logic

If you use Condition or Condition Agent nodes:

Verify each output path triggers correctly
Test boundary conditions
Confirm "else" cases work as expected

Test Interactive Elements

For Buttons and Leads nodes:

Click all buttons to verify behavior
Submit lead forms with valid/invalid data
Test on mobile devices if using website integration

Test Cross-Channel

If deployed to multiple channels:

Test on website (bubble and iframe if applicable)
Test on each messaging platform
Verify consistent behavior across channels

Keep the default Condition → Welcome flow
Replace default welcome message with branded greeting
Update Buttons to reflect your top 3 use cases
Add a Leads node after buttons for early contact capture

FAQ Automation

Scenario: Automatically answer common questions without AI generation

Implementation:

Add a Condition Agent node after Start
Configure it to detect FAQ topics (returns, shipping, pricing, etc.)
Create a dedicated Answer node for each topic
Connect Condition Agent outputs to respective Answer nodes
Add "catch-all" path to Knowledge → Answer Generator for unknown questions

Lead Capture with Qualification

Scenario: Collect leads and qualify them before routing

Implementation:

After initial greeting, show Buttons asking "How can we help?"
Each button path leads to relevant information (Answer node)
Add Leads node: "Want us to reach out? Leave your info"
After lead submission, show thank you message and set expectations

Sentiment-Based Routing

Scenario: Detect frustrated users and provide priority support

Implementation:

Add Condition Agent node analyzing sentiment
Negative sentiment output → Empathetic Answer + Leads form
Neutral/Positive output → Standard Knowledge → Answer Generator flow
Ensures frustrated customers get immediate attention

Best Practices

Design Principles

Start Simple

Begin with the default workflow
Add complexity gradually
Test after each change

Keep It Logical

Organize nodes left-to-right (start → end)
Group related nodes together
Use clear, descriptive node labels

Avoid Dead Ends

Every path should lead somewhere meaningful
Use Answer nodes to close conversations gracefully
Consider adding "Anything else?" loops

Organization Tips

Use Visual Layout:

Horizontal flows are easier to follow than vertical
Space nodes consistently
Align connected nodes for clarity
Use the snap-to-grid feature for clean layouts

Label Everything:

Rename nodes from defaults ("Answer Node 1" → "Welcome Message")
Use descriptive Condition labels ("Check if returning user")
Comment complex logic

Template Best Practices:

Save workflow milestones as templates
Document template purposes in descriptions
Version template names ("FAQ Flow v2")
Test templates thoroughly before saving

Performance

Minimize Complexity:

Avoid deeply nested conditions (limit to 3-4 levels)
Use efficient Condition logic (exact matches over regex when possible)
Limit HTTP requests to essential calls only

Optimize Knowledge Searches:

Assign only relevant training content to Support BVs
Keep training data well-organized
Remove outdated content regularly

Week 1: Explore Default
- Understand the default workflow
- Test with various questions
- Observe how routing works
Week 2: Simple Modifications
- Update welcome messages
- Customize button text
- Adjust Answer node messages
Week 3: Add Complexity
- Add a new Condition node
- Create a custom FAQ path
- Experiment with templates
Week 4: Advanced Features
- Build a lead capture flow
- Try Condition Agent for intelligent routing
- Create and save your first template

Getting Help

Getting Started Tour: Click the rocket icon for guided walkthrough
Quick Help: Click the info icon for canvas shortcuts
Documentation: Refer back to this guide
Support: Contact support for workflow design assistance

Action Map

Nodes won't connect

Changes not showing in test chatbot

Published version not updating

Template import not working

Workflow execution errors

Knowledge not finding relevant content

Buttons or Leads not displaying

Configure Personality

AI Settings

Add Training Data

Test Your Support BV

On this page