​
Understanding Branches in Atoms

In Atoms, a Branch is a connector that links one Node to another within a conversation workflow. It determines the flow of the conversation, directing which node should be activated next based on user input or predefined conditions.

​
Example: Restaurant Reservation Booking Workflow

Continuing from our Restaurant Reservation Assistant, branches define how the conversation moves forward based on the customer’s responses.
  1. Greet & Ask for Reservation Intent
    • Branch 1: If the user declines, transition to the End Call Node.
    • Branch 2: If the user wants to book, move to Ask for Date & Time.
  2. Ask for Date & Time
    • Branch: Continue to Collect Guest Details.
  3. Collect Guest Details
    • Branch: Collects the Guest details such as Name and Number of guests and proceeds to Send Reservation Request via the Webhook Node.
  4. Send Request (Webhook Node)
    • Branch: If successful, transition to Confirm & End Call.
  5. Confirm & End Call
    • Branch: Conclude the conversation.

​
Creating a Branch in Atoms

To modify or create branches in your Restaurant Reservation Agent, simply click to edit branches and define the conditions for transitioning from one node to another. Edit Branch

​
Parameters of a Branch

When setting up a branch, you can choose between two different types of decision-making logic by selecting the appropriate tab: Branch configuration interface

​
Prompt Tab (AI Decision)

Use the Prompt tab when you want the AI to decide which path to take based on conversation context.
ParameterDescription
Branch PromptA short decision label describing when this branch should be followed.
Branch DescriptionA more detailed explanation of the condition or reason for this transition. Helps improve response accuracy.

​
Condition Tab (Variable Logic)

Use the Condition tab when you want to make decisions based on specific variable values extracted from the conversation.
ParameterDescription
Variable NameThe name of the variable to check.
OperatorThe comparison operator (==, !=, >, <, >=, <=, contains, not_contains, is).
ValueThe exact value to compare against.

​
When to Use Each Type

​
Use Prompt-Based When:

  • Decisions require natural language understanding
  • Multiple factors need to be considered
  • The logic is conversational or nuanced
  • You want the AI to make context-aware decisions

​
Use Condition-Based When:

  • Working with specific numeric values
  • Checking exact string matches
  • Need deterministic, rule-based routing
  • Variables have been extracted from previous nodes

​
Examples

​
Prompt-Based Example

Node: Check Restaurant Availability
Branch 1 (Prompt):
Prompt: “Restaurant is full”
Description: “Choose this path when customer mentions the restaurant is fully booked” Branch 2 (Prompt):
Prompt: “Table available”
Description: “Choose this path when a suitable table is available for booking”

​
Condition-Based Example

Node: Check Table Size
Branch 1 (Condition):
Variable: guest_count
Operator: >
Value: 8
→ Routes to “Large Group Handling” Branch 2 (Condition):
Variable: guest_count
Operator: <=
Value: 8
→ Routes to “Normal Booking Process”

​
Best Practices

  1. Clear Variable Names: Use descriptive variable names in condition-based branches
  2. Detailed Descriptions: Provide clear context in prompt-based descriptions
  3. Variable Validation: Ensure variables exist before using them in conditions

​
Wrapping Up

Branches form the backbone of conversational flow, ensuring smooth transitions between nodes. With the choice between AI-driven decisions (Prompt) and variable-based conditions, you can create more dynamic and intelligent workflows tailored to your specific needs. In the next section, we’ll explore the different types of nodes you can add and configure.