> ## Documentation Index > Fetch the complete documentation index at: https://docs.subverseai.com/llms.txt > Use this file to discover all available pages before exploring further. # Voice Agent Node > Make outbound calls to customers with AI voice agents ## Overview The Voice Agent Node enables your workflow to make outbound calls to customers using your configured AI voice agents. This is the core action block for customer outreach, follow-ups, and automated calling campaigns. ## Use Cases * **Outbound Campaigns**: Launch sales or marketing call campaigns * **Appointment Reminders**: Automated reminder calls for appointments * **Order Confirmations**: Confirm orders and delivery details * **Payment Reminders**: Follow up on pending payments * **Customer Surveys**: Collect feedback and ratings * **Lead Follow-ups**: Call leads from your CRM * **Event Notifications**: Inform customers about events * **Service Updates**: Notify about service changes or outages ## Inputs The Voice Agent Node accepts the following inputs, which can be linked from previous blocks or manually configured: ### **Bot\_Number** (Required) * Phone number the voice agent will use to make the call * Can be linked from a previous block or selected from dropdown * Data linkage overrides dropdown selection * Format: +\[country code]\[number] **Example:** `+91xxxxxxxxxx` ### **Use\_Case** (Required) * Defines which voice agent will handle the call * Can be linked from a previous block or selected from dropdown * Data linkage overrides dropdown selection * Must match an existing agent in your account **Example:** `customer_support`, `sales_agent` ### **Customer\_Number** (Required) * Customer's phone number to call * Format: +\[country code]\[number] * Must be a valid, callable number **Example:** `+91xxxxxxxxxx` ### **Customer\_Details** * Customer information in JSON format * Used to personalize the conversation * Can include: name, email, order ID, account details, etc. **Example:** ```json theme={null} { "name": "John Doe", "email": "john@example.com", "order_id": "ORD-12345", "product": "Premium Plan", "amount": "₹2,499" } ``` ### **Schedule\_Time** * Schedule the call for a specific future time * Format: ISO 8601 date format `YYYY-MM-DDTHH:MM:SSZ` * Default: Current time (immediate execution) **Example:** `2025-12-10T15:30:00Z` ### **Agent\_Options** * Override default agent settings for this specific call * Allows per-call customization * Can modify: language, voice, initial prompt, etc. **Example:** ```json theme={null} { "language": "hi-IN", "voice": "Riya", "initial_prompt": "Hello, this is a reminder about your appointment tomorrow." } ``` ### **No\_Of\_Retries** * Number of retry attempts if the call fails * Default: 0 (no retries) * Recommended: 2-3 for important calls **Example:** `2` ### **Retries\_After\_In\_Hrs** * Time interval between retry attempts (in hours) * Works with No\_Of\_Retries * Recommended: 1-2 hours **Example:** `1` ### **Start\_Working\_Hour** * Start of business hours in 24-hour format (HH:MM) * Calls scheduled before this time will be rescheduled * Respects customer preferences **Example:** `09:00` ### **End\_Working\_Hour** * End of business hours in 24-hour format (HH:MM) * Calls after this time will be rescheduled to next day * Prevents late-night calls **Example:** `18:00` ### **Timezone** * Timezone for working hours calculation * Default: Asia/Kolkata * Important for international customers **Example:** `America/New_York`, `Europe/London`, `Asia/Kolkata` ## Outputs After the call is completed, the Voice Agent Node provides the following outputs: ### **Bot\_Number** * Phone number the voice agent used to make the call * Same as input value ### **Use\_Case** * Voice agent that handled the call * Same as input value ### **Customer\_Number** * Customer's phone number that was called * Same as input value ### **Customer\_Details** * Customer information that was provided * Same as input value ### **Agent\_Options** * Options that were used to override settings * Same as input value (if provided) ### **Call\_Time** * Timestamp when the call was made * Format: ISO 8601 `YYYY-MM-DDTHH:MM:SSZ` **Example:** `2025-12-10T10:30:00Z` ### **Call\_Status** * Final status of the call * Possible values: * `connected`: Call was successfully connected * `couldn't connect`: Call failed to connect **Example:** `connected` ### **Call\_Duration** * Duration of the call in seconds * Excludes ring time * 0 for failed connections **Example:** `125` (2 minutes 5 seconds) ### **Call\_Direction** * Direction of the call * Always `outbound` for this node **Example:** `outbound` ### **Transcript** * Complete transcript of the call conversation * Includes speaker tags (Assistant vs Customer) * Available after call completion **Example:** ``` Assistant: Hello John, this is a reminder about your appointment tomorrow at 3 PM with Dr. Smith. Customer: Oh yes, thank you for reminding me. Assistant: You're welcome! Do you need directions to the clinic? Customer: No, I know where it is. Thank you! Assistant: Great! We'll see you tomorrow. Have a nice day! ``` ### **Analysis** * Post-call analysis data (if configured) * Extracted information from the conversation * Custom fields defined in agent settings * Type: String ### **Call\_Recording\_Url** * URL to access the call recording * Available after call completion * Can be used for quality assurance * Type: String ### **Call\_Id** * Unique identifier for the call * Used for tracking and reference * Type: String ### **No\_Of\_Retries** * Number of retries configured (if provided) ### **Retries\_After\_In\_Hrs** * Retry interval configured (if provided) ### **Start\_Working\_Hour** * Working hours start time (if provided) ### **End\_Working\_Hour** * Working hours end time (if provided) ### **Timezone** * Timezone used for scheduling (if provided) ### **Call\_Priority** * Priority level configured (if provided) * Type: Number ### **Call\_On\_Date** * Scheduled date configured (if provided) * Format: `YYYY-MM-DD` ## Configuration Examples ### Basic Outbound Call ``` Bot_Number: +91xxxxxxxxxx Use_Case: sales_agent Customer_Number: +91xxxxxxxxxx Customer_Details: {"name": "John Doe"} ``` ### Scheduled Call with Retries ``` Bot_Number: +91xxxxxxxxxx Use_Case: appointment_reminder Customer_Number: +91xxxxxxxxxx Customer_Details: {"name": "Jane Smith", "appointment": "Dec 10, 3 PM"} Schedule_Time: 2025-12-09T10:00:00Z No_Of_Retries: 2 Retries_After_In_Hrs: 2 ``` ### Call with Working Hours ``` Bot_Number: +91xxxxxxxxxx Use_Case: payment_reminder Customer_Number: +91xxxxxxxxxx Customer_Details: {"name": "Bob Johnson", "amount": "₹5,000"} Start_Working_Hour: 09:00 End_Working_Hour: 18:00 Timezone: Asia/Kolkata ``` ### Call with Custom Agent Options ``` Bot_Number: +91xxxxxxxxxx Use_Case: customer_support Customer_Number: +91xxxxxxxxxx Customer_Details: {"name": "Alice Brown"} Agent_Options: { "language": "hi-IN", "voice": "Riya", "initial_prompt": "नमस्ते, मैं आपकी कैसे मदद कर सकती हूं?" } ``` ## Workflow Integration ### Connecting with Other Nodes **From Trigger Nodes:** * Call API Node → Voice Agent Node * Upload Data Node → Voice Agent Node * Incoming Call Node → Voice Agent Node (for callbacks) **To Function Nodes:** * Voice Agent Node → Send Email Node (follow-up email) * Voice Agent Node → Send WhatsApp Node (confirmation message) * Voice Agent Node → Webhook Node (CRM update) * Voice Agent Node → Schedule Callback Node (if needed) **With AI Processing:** * Voice Agent Node → AI Query Agent Node (analyze call outcome) * AI Query Agent Node → Voice Agent Node (make decision-based calls) ## Best Practices ### Call Timing * **Respect Time Zones**: Always set appropriate timezone * **Business Hours**: Use Start/End Working Hour constraints * **Spacing**: Don't schedule all calls at once * **Retry Strategy**: Set reasonable retry intervals ### Personalization * **Use Customer Data**: Include name and relevant details * **Context**: Provide conversation context in Customer\_Details * **Dynamic Prompts**: Use Agent\_Options for per-call customization * **Relevant Information**: Only include necessary data ### Quality Assurance * **Test First**: Make test calls before launching campaigns * **Monitor**: Check Call History regularly * **Review Transcripts**: Listen to sample calls * **Iterate**: Improve based on feedback ### Performance * **Batch Processing**: Use Upload Data Node for bulk campaigns * **Rate Limiting**: Don't exceed your concurrent call limit * **Error Handling**: Set up retry logic for failed calls * **Monitoring**: Track success rates ## Monitoring Call Performance ### Key Metrics to Track **Connection Rate** * Percentage of calls that successfully connected * Target: >80% * Low rate indicates phone number or timing issues **Average Call Duration** * Typical length of conversations * Helps identify engagement levels * Very short calls may indicate problems **Completion Rate** * Calls that reached intended outcome * Based on post-call analysis * Indicates agent effectiveness **Retry Success** * Percentage of retries that succeeded * Validates retry strategy * Adjust timing if low ### Analyzing Results Use the outputs to: * **Segment Customers**: Group by call outcome * **Trigger Actions**: Based on call status * **Update Records**: Send data to CRM * **Schedule Follow-ups**: For unsuccessful calls * **Generate Reports**: Export for analysis ## Troubleshooting ### Call Not Connecting **Possible Causes:** * Invalid phone number format * Number is unreachable or blocked * Telephony provider issues * Insufficient credits **Solutions:** * Verify phone number format (+\[country]\[number]) * Test with a known working number * Check telephony integration status * Review account balance ### Short Call Duration **Possible Causes:** * Customer hung up immediately * Poor audio quality * Agent not responding * Wrong number **Solutions:** * Review greeting message * Test audio quality * Check agent configuration * Validate target audience ### Calls Not Scheduling **Possible Causes:** * Invalid Schedule\_Time format * Time in the past * Working hours conflict **Solutions:** * Use ISO 8601 format * Ensure future timestamp * Review working hour settings ## Next Steps Process call outcomes with AI Schedule follow-up calls Send post-call emails Monitor call performance Link Customer\_Details from previous blocks to create dynamic, personalized calls based on real-time data. Always comply with local telemarketing regulations and respect do-not-call lists. Obtain proper consent before making automated calls.