DialNexa Voice AI: Dynamic Variables and Fallback Settings
This video explains how to use Dynamic Variables within the DialNexa platform to personalise Voice AI conversations at scale. Dynamic Variables enable AI agents to automatically insert lead-specific information such as names, colleges, degrees, graduation years, company names, or any other custom attributes into conversations, allowing businesses to deliver highly personalised experiences across thousands of outbound calls without manually modifying individual agent prompts.
The tutorial begins by explaining the need for Dynamic Variables when running large-scale outbound Voice AI campaigns. Instead of creating separate AI agents or editing prompts for every individual lead, businesses can define reusable placeholders that are automatically populated with data from their lead database or CSV file. This makes it possible to deploy AI sales agents, lead qualification workflows, recruitment assistants, customer support agents, appointment booking systems, and enterprise Conversational AI solutions across thousands of contacts while maintaining personalised conversations for every recipient.
The video demonstrates how Dynamic Variables can be used inside the Welcome Message to greet each lead by name. Rather than hardcoding names into the prompt, the agent uses placeholders enclosed within double curly braces, such as `{{lead_name}}`, which are automatically replaced with the corresponding value mapped to each phone number in the uploaded contact list. As the Voice AI agent moves from one call to the next, these values dynamically change based on the lead information associated with every contact.
The tutorial also explains the correct syntax for creating Dynamic Variables, including the use of double curly braces and alphanumeric variable names without spaces. Users learn how these variables map directly to column names in their CSV files, ensuring that every lead attribute is correctly associated with the intended phone number during bulk calling campaigns. This flexible approach allows businesses to personalise conversations using multiple variables simultaneously, including names, educational background, organisations, job titles, graduation years, locations, or any other custom fields relevant to their use case.
The video further demonstrates how multiple Dynamic Variables can be combined within a single sentence to create highly contextual conversations. For example, an AI agent can reference a prospect's name, college, degree, and graduation year within the same dialogue, making interactions feel significantly more natural, relevant, and human-like. These variables can be used throughout both the Welcome Message and the Agent Prompt, enabling consistent personalisation across every stage of the conversation.
Another important concept covered in the tutorial is Fallback Settings for Dynamic Variables. Since real-world datasets are often incomplete, businesses may occasionally have missing values for certain lead attributes. Instead of allowing the Large Language Model (LLM) to hallucinate or generate incorrect information, DialNexa allows users to configure default fallback values for each Dynamic Variable. These defaults are automatically substituted whenever a corresponding value is missing from the uploaded lead data, helping maintain conversation quality and preventing inaccurate responses.
The tutorial also shares practical best practices for configuring fallback values. Rather than assigning potentially incorrect information such as random names, users can choose fallback values that naturally fit the conversation. For example, when a lead's name is unavailable, using a simple placeholder such as a period allows the LLM to intelligently complete the sentence with phrases like "Am I speaking with you?" instead of inventing a name. This approach creates a far more natural customer experience while reducing hallucinations and improving overall conversation quality.