​
Introduction to Workflows

Workflows in Welco AI are visual, node-based automation systems that allow you to create sophisticated conversational experiences for your AI receptionist. Instead of writing complex code, you can drag and drop nodes on a visual canvas to build interactive call flows that handle everything from simple greetings to complex business processes.

​
What Are Workflows?

intro to voice agents workflows Workflows are structured sequences of actions that define how your AI receptionist handles different scenarios during phone calls. Each workflow consists of interconnected nodes that represent different types of actions - conversations with callers, function calls to external systems, call transfers, and more. Think of workflows as conversation maps that guide your AI through various paths based on caller responses, business rules, or external data. They provide a visual way to design complex call handling logic without requiring programming knowledge.

​
Core Concepts

​
Visual Node-Based System

Workflows use a drag-and-drop canvas where you connect different types of nodes to create call flows: Workflownodes Pn
  • Scenario Node: Defines when the workflow triggers (always the starting point)
  • Conversation Nodes: Handle AI interactions with callers
  • Function Nodes: Execute business logic or integrate with external systems
  • Logic Split Nodes: Create conditional branching based on data or responses
  • Transfer Nodes: Route calls to phone numbers or other departments
  • End Call Nodes: Terminate the conversation (endpoint nodes)

​
Connection-Based Flow Control

Nodes connect through handles (connection points) that define the conversation flow:
  • Each node can have incoming connections (targets) and outgoing connections (sources)
  • Scenario nodes only have outgoing connections (starting points)
  • End Call nodes only have incoming connections (endpoints)
  • Most nodes have multiple outgoing handles based on possible outcomes

​
Variable and Context Management

Workflows maintain context throughout the conversation:
  • Variables store information collected during the call
  • Session data persists across different nodes in the same call
  • Global settings apply across all workflows for an agent

​
When to Use Workflows vs Simple Responses

​
Use Workflows When You Need:

Complex Decision Trees
  • Multi-step qualification processes
  • Different paths based on caller responses
  • Conditional logic that varies by caller data
External System Integration
  • CRM lookups during calls
  • Appointment scheduling with calendar systems
  • Database queries for customer information
  • Third-party API calls for real-time data
Multi-Modal Interactions
  • Combining voice conversations with SMS follow-ups
  • Collecting structured information through forms
  • Transferring calls based on complex criteria
Business Process Automation
  • Lead qualification and scoring
  • Appointment booking with availability checking
  • Order processing with inventory validation
  • Support ticket creation and routing

​
Use Simple Responses When You Have:

Straightforward Questions
  • Basic FAQ responses
  • Simple information requests
  • Standard greetings or closings
Static Information
  • Business hours
  • Location details
  • General pricing information
Single-Step Interactions
  • Quick confirmations
  • Basic yes/no questions
  • Simple data collection

​
Workflow Execution Model

​
Template vs Instance

Workflow Templates are the designs you create - they define the structure, nodes, and connections but don’t execute directly. Workflow Instances are created each time a template runs during an actual call. Each phone call creates a new instance that maintains its own variables and state.

​
Execution Engine

When a call comes in, the system:
  1. Evaluates Scenarios - Determines which workflow template matches the call context
  2. Creates Instance - Spawns a new workflow instance from the template
  3. Executes Nodes - Processes each node in sequence based on connections
  4. Maintains State - Tracks variables and context throughout the call
  5. Logs Everything - Records execution details for debugging and analytics

​
Getting Started

To begin with workflows:
  1. Plan Your Flow - Map out the conversation paths on paper first
  2. Start with Scenario - Every workflow begins with a scenario node
  3. Build Step by Step - Add nodes one at a time and test connections
  4. Configure Settings - Set up node-specific configurations
  5. Test Thoroughly - Use the execution logs to verify behavior
  6. Iterate and Improve - Refine based on real call data and feedback
Best Practice: Start with simple workflows and gradually add complexity. It’s easier to debug and maintain smaller, focused workflows than large, monolithic ones.

​
Next Steps

Now that you understand the fundamentals, explore: Workflows transform your AI receptionist from a simple question-answering bot into an intelligent business process automation tool that can handle complex scenarios while maintaining natural conversation flow.