VocaLoop

Documentation

VocaLoop is mid-call data capture for AI voice agents. When a call needs something a voice can’t carry, a correctly spelled email, a photo of an ID, a signature, the agent sends a form to the caller’s phone and the answers flow back while they’re still on the line.

The same loop works outside calls: any system that can make an HTTP request can send one, track it, and receive the response. These docs cover setup, channels, billing, and the API.

A loop, end to end
  1. Agent“I need your email and a photo of your driver’s licence.”
  2. WhatsAppHi Ravi, Acme needs a few details. Fill them here: vocaloop.link/f/8k2…
  3. Response{ "status": "submitted", "email": "ravi@acme.com", "licence": "uploaded" }

How a loop works

  1. 01Ask

    Your agent or code states what it needs in one sentence.

  2. 02Generate

    The AI builder turns it into a branded form with the right field types, validation, and uploads.

  3. 03Deliver

    A single-use link goes to the caller over WhatsApp, SMS, or email using your template.

  4. 04Return

    The answers come back by polling the loop or through a webhook, with per-loop analytics.

Start here

Building a voice agent

Wire VocaLoop into Vapi, Retell, ElevenLabs, or LiveKit as one tool definition.

Getting Started →
Integrating from code

Two endpoints and a project API key. Add a webhook to be pushed instead of polling.

Secure Link Loop API →
Trying it by hand

Send a real form to your own phone from the Playground before you build anything.

Playground & Test Sends →
Project
One use case: its form, templates, branding, API keys, and responses.
Loop
One round trip, from prompt to submitted answers, tracked by a request id.
Fill link
The unguessable, expiring URL the recipient opens. The first submission closes it.
Request
A loop's lifecycle on the Responses page: pending, submitted, failed, or expired.