Hi, I need urgent guidance on a project we are doing for a hackathon. We have a fairly long web form that a user needs to fill, and a voice agent is supposed to guide them through the form. The form has multiple pages and each page has between 5 to 10 fields, with each field having a unique identifier. Voice agent is supposed to ask questions one by one , gather the answers, map the answers to the form field ids - most important, and save it in structured data. So far using the questionnire in the prompt has been mostly a miss, it would either miss questions or not map the correctly to the form field Ids. Whats the best way to solve this on Vapi which is reliable ?

Use the Vapi Workflows "Gather" node to reliably map answers to form field IDs. Define each form field as a variable in the Gather node, matching your field IDs, and set their types and required status. For example:

json
{
  "variables": [
    { "name": "email", "description": "User's email address", "type": "string", "required": true },
    { "name": "age", "description": "User's age", "type": "number", "required": true }
  ]
}

This ensures each answer is collected and mapped to the correct field, with structured data output for your form. Source: - [Vapi Workflows Gather Node Documentation](https://docs.vapi.ai/workflows/nodes/gather)

The above link throws 404

Define a clear structuredDataSchema, use a focused structuredDataPrompt, ask questions step-by-step, and validate results using Vapi’s call analysis tools..