Executing Process Action
Symphona Converse’s Agent Editor allows you to easily execute processes using our actions. You are able to execute any Flow Process you have created to be used through the AI Agent.
In the following sections, we will cover additional fields to help you maximize the capability of your agent!
Confirm User Inputs Before Executing
Within the Process Configuration section, the "Confirm User Inputs Before Executing" toggle gives you control over whether the agent should require explicit user confirmation before proceeding with an action.
When this confirmation toggle is enabled, the agent will present the collected input parameters to the user and wait for their approval before executing the action, providing an additional layer of verification to ensure accuracy. When disabled, the agent will proceed directly with action execution once all required parameters are gathered, streamlining the process for scenarios where immediate execution is preferred. This setting is particularly useful for actions that have significant consequences or when you want to give users a final opportunity to review their inputs before the system processes their request.
Dynamic Parameter Validation
The dynamic parameter validation enables you to extract and validate specific values from user input while maintaining conversational context. You define validation rules that describe exactly what format or content is required (such as "Must be a valid phone number with country code" or "Must include house number and street"), and Symphona will intelligently extract relevant values from user responses and check them against your criteria. The system is designed to be strict about validation while being smart about context - it will only switch topics when users explicitly request a change, ensuring that incorrectly formatted data doesn't accidentally derail the validation process.
To use this validation effectively, write clear and specific validation rules that describe both the required format and any mandatory components. The system works best when your rules are precise about what constitutes a valid response, as it will provide helpful feedback when validation fails, telling users exactly what's missing or incorrect and how to fix it. For example, if validating an address, specify whether you need just a street name or require both house number and street name. The AI will automatically handle edge cases like partial responses or completely unrelated input, only switching context when users explicitly signal they want to discuss something else, making it ideal for forms, data collection workflows, or any scenario where you need to ensure users provide properly formatted information before proceeding.
Store Process Execution Parameters
Within the Post-Action Behavior section you can store process execution parameters in agent memory when triggering actions within a converse agent. You can toggle this feature on or off and choose whether to store individual parameters selectively or store all parameters automatically. When storing individual parameters, you can add specific parameters by defining their name and purpose, helping the agent remember important information from completed actions for use in future conversations.
For example, you might store a parameter called "CustomerID" with the purpose "Customer identifier retrieved during account lookup" so the agent can reference the specific customer information in subsequent interactions without requiring the user to provide their details again. Each parameter includes a name field for identification and a purpose field to describe what information it contains, with character limits to keep descriptions concise. You can add multiple parameters using the Add Parameter button and remove them individually using the delete option, giving you precise control over what information the agent retains from each action execution.
Conversation Metadata Fields
Conversation metadata fields enable you to dynamically populate action parameters with contextual information from ongoing conversations when triggering actions such as API requests, process executions, or service ticket creation. You can configure action parameters in the generative Agent Editor to automatically include relevant conversation data by using special keywords that are replaced with actual values at runtime.
The system provides several predefined metadata keywords that you can insert into static action parameter fields, including SESSION_USER_ID, SESSION_USER_FIRST_NAME, SESSION_USER_LAST_NAME, SESSION_USER_PHONE, SESSION_USER_EMAIL, SESSION_ID, SESSION_USER_LOCATION, SESSION_START_TIME, and SESSION_AGENT_ID. These keywords allow you to seamlessly pass user information, session details, and conversation context to your actions without requiring users to re-enter information they've already provided during the conversation.