Aurra · Imitate Labs

AI Telecaller: Voice Agents That Call Every Lead

Upload your leads and pick an AI voice. Aurra calls every one of them in their own language, tries again when nobody picks up, and tags each call with how it went.

Outbound calls: 580 calls, 84% connected, recordings per call
Dashboard: active campaigns, leads, credits, missed callbacks

Platforms

  • Web
  • Voice

Overview

Aurra, from Imitate Labs, gives a business an AI team that calls and chats with its customers. The AI Telecaller is the part that calls. You upload leads or connect a lead source and choose a voice persona. Aurra then rings every lead, talks to them in their language, retries anyone who didn't answer, and tags each call with a business outcome. We built the whole product, including the self-hosted voice pipeline that runs in real time.

Project details

Client
Imitate Labs (product: Aurra)
Industry
AI / SaaS: sales and support automation for Indian businesses (real estate, education, healthcare, lending…)
Market
India (11 languages)
Engagement
Jan 2026 – present (Telecaller since May 2026)
Our role
All of it: product, voice AI pipeline, WhatsApp bot, CRM, billing, cloud
Platform
Web app + self-hosted real-time voice bot

What we built

  • Campaigns from a CSV upload, a webhook / API, Meta lead ads, or inbound numbers
  • AI voice personas with versioned prompts, so changes go live on the next call with no redeploy
  • Voice cloning: upload a sample and use a custom voice on every campaign
  • 11 languages, including Indian-language voices
  • Live call pipeline: who is on a call now, who is next, automatic retries and scheduled callbacks
  • Auto dispositions: after each call an AI reads the transcript and marks Interested, Callback, Appointment Fixed, Not Interested… plus each business's own custom outcomes
  • Call recordings and transcripts, missed-callback and inbound-call handling
  • Mid-call tools: look up the knowledge library, book appointments, send a WhatsApp during the call
  • Developer API, API playground, and API campaigns where each lead brings its own script
  • Test calls before going live

Inside the product

Dashboard: active campaigns, leads, credits, missed callbacks
Outbound calls: 580 calls, 84% connected, recordings per call
Campaigns with status, usage and test call
New campaign: start time, calling hours and automatic retries
Call outcomes (dispositions) the AI decides from the conversation
Inbound calls answered by the AI
Missed callbacks to follow up
Edit campaign: retries and delays
Aurra landing page
01 / 09

Dashboard. Active campaigns, leads, credits, missed callbacks

Technology

Tech stack

  • Python
  • Pipecat (real-time voice pipeline, self-hosted)
  • Google Gemini (with prompt caching)
  • Cartesia voice cloning
  • Sarvam (Indian languages)
  • Acefone telephony
  • NestJS
  • PostgreSQL

Want a product like this?