Real-Time Web Form Filling Voice Agent

A voice agent that conducts phone questionnaires while automatically filling out web forms in real-time using browser automation. The agent collects answers through natural conversation and asynchronously fills form fields in the background, maintaining smooth conversation flow without blocking.

Try the demo View source Fork template

The numbers

latency—

cost / min—

frameworkCartesia

The stack

telephonyWeb Only

speech-to-textCartesia STT

llmGemini 2.5 Pro

text-to-speechCartesia Sonic-3

System prompt

raw_prompt

Hello! I'm here to help you fill out an application form today...

Config

config.json

{
  "tools": [
    {
      "name": "start_questionnaire",
      "description": "Starts the questionnaire flow and asks the first question. This is a passthrough tool that directly emits AgentSendText events for deterministic conversation flow."
    },
    {
      "name": "record_form_field",
      "description": "Records an answer and fills the corresponding form field in the browser. Asynchronously fills the form in the background while immediately asking the next question. This is a passthrough tool for predictable conversation flow."
    }
  ],
  "form_url": "configurable",
  "model_id": "gemini-pro",
  "auto_submission": true,
  "async_form_filling": true,
  "browser_automation": "Stagehand + Browserbase",
  "eager_browser_initialization": true
}

Tools

start_questionnaireStarts the questionnaire flow and asks the first question. This is a passthrough tool that directly emits AgentSendText events for deterministic conversation flow.
record_form_fieldRecords an answer and fills the corresponding form field in the browser. Asynchronously fills the form in the background while immediately asking the next question. This is a passthrough tool for predictable conversation flow.

Voice AI recipes, picks, and analysis.

Get the useful new templates plus the occasional teardown of what’s working in production voice AI.