Quickstart

This walkthrough takes an empty account to a graded call and a live integration. The free Starter plan includes 300 call minutes a month, which is plenty for it.

1. Create a workspace

1. Sign up and verify your email

   Go to voicelens.app/auth/signup. The verification link is valid for 24 hours.

2. Name your workspace

   A workspace holds calls, channels, templates, members and keys. If you serve several clients or brands, plan on one workspace per client; you can add more later.

2. Grade a first call

1. Open Upload

   In the dashboard, choose Upload and drop in a recording. Supported formats are mp3, wav, m4a, ogg, webm and flac, up to 500 MB per file.

2. Pick a rubric (or let VoiceLens draft one)

   A fresh workspace has no rubric yet. Either start from one of the built-in templates (real estate, attorney intake, telecom and ISP support) under Templates, or upload anyway: on the first call in a rubric-less workspace, VoiceLens drafts a rubric from the transcript for you to review. Drafts count against your monthly AI rubric allowance.

3. Wait for processing

   A call moves through Pending, Transcribing, Analyzing and Completed. Short calls usually finish in under a minute.

4. Open the scorecard

   Click the call to see its health score, faults, latency, slots and sentiment. Reading a scorecard explains each panel.

3. Connect your phone system

Recordings should arrive on their own once you go live. Create a channel for each line, bot or queue and pick one of three ways to send audio.

1. Create a channel and an ingest token

   Under Channels, create a channel, assign it a rubric template and create an ingest token. The token (prefix vk_) is shown once; store it in your secret manager. See Channels and ingest tokens.

2. Send audio after each call

   The simplest integration is one request per finished call:

   curl -X POST "https://voicelens.app/api/ingest/$CHANNEL_ID/audio/from-url" \
     -H "X-Ingest-Token: $VK_TOKEN" \
     -H "Content-Type: application/json" \
     -d '{
       "url": "https://recordings.example.com/calls/8f2a.mp3",
       "externalCallId": "CA8f2a",
       "metadata": { "agentId": "bot-v12", "agentType": "ai" }
     }'

   Multipart uploads and pre-signed upload URLs are also available; the ingest API reference covers all three.

3. Or point Twilio at VoiceLens

   If your calls run on Twilio, set the channel provider to Twilio, enter your Account SID and Auth Token, and use https://voicelens.app/api/ingest/{channelId}/twilio/recording-status as the recording status callback. Completed recordings are fetched and graded automatically.

4. Get results into your tools

• Webhooks deliver conversation.completed and threshold breaches to your QA queue with an HMAC signature. See Webhooks.
• API keys (prefix vlk_) give read-only access to conversations and analytics for your own reporting. See Authentication.
• Alerts and the weekly digest email your team when average health drops or failures rise. See Alerts and digest.
• Embedded dashboards put a themed read-only view inside your own portal. See Embedded dashboards.

5. Calibrate before you rely on it

Grade ten or twenty calls by hand, mark them as a golden set with the score you expect, and run calibration against your rubric. The report tells you how far the automated grade is from your reviewers before you use it for coaching or compliance decisions.