ABOUT CALLMANGAL

Building the voice layer for business conversations.

CallMangal is building AI voice agents for businesses that need to communicate with customers at scale — across sales, support, qualification, bookings and follow-ups. India first. Built to scale globally.

OUR MISSION

Make intelligent voice automation accessible to every business — without requiring every business to become an AI engineering company.

Most companies do not need to build a voice stack. They need a working agent that answers the phone properly, in their customers' language, and hands over to a person when it should.

OUR STORY

Built from other businesses' problems.

CallMangal is a product operated by START DESIGNS.

START DESIGNS began by building MVPs and web and mobile products for other founders. EcomHeroes followed, designing and developing Shopify stores for ecommerce brands. Years of sitting that close to other companies' operations surfaced the same gap again and again: everyone had a website, a CRM and an ad budget — and nobody had solved the phone.

Calls rang out after hours. Leads went cold before anyone called back. A business was always one busy afternoon away from losing a customer it had already paid to acquire.

CallMangal was created around a simple observation: business communication in India is multilingual, high-volume and still heavily dependent on phone conversations. The tooling available either assumed an engineering team, or assumed a call centre.

So CallMangal is being built to make AI voice technology practical for the people who actually field those calls — businesses, agencies, resellers and service providers — with a visual way to build agents, multilingual workflows, inbound and outbound calling, browser-based testing before anything goes live, and a white-label model for partners who want to deliver it under their own brand.

Every capability exists because a real business asked for it first, not because it demonstrated well.

WHY INDIA FIRST

A demanding place to build voice AI.

Not a market to serve later — a proving ground that makes the technology better.

Many languages, in one call

Customers switch between languages mid-sentence. An agent that only handles clean, single-language speech does not survive a real Indian phone call.

Accents and audio conditions

Regional accents, background noise and variable line quality are the normal case, not the edge case. Building for that raises the floor everywhere.

Phone-first customers

A large share of customer interaction still starts with a call. The phone is not a legacy channel here; for many businesses it is the primary one.

A large SMB and mid-market base

Enormous numbers of businesses handle serious call volume without a call-centre budget or an in-house engineering team.

Tooling has to be accessible

If configuring an agent requires an AI engineer, most of the market is excluded. That constraint shaped the visual builder more than anything else.

Hard here, portable elsewhere

India-first engineering is a demanding environment. Technology that holds up in it is well placed to scale to other markets.

LEADERSHIP & TEAM

The people building CallMangal.

A product team spanning voice engineering, AI, product and go-to-market.

Leadership

Vipin Agarwal
Founder & CTO

Product vision, architecture and technology strategy.

LinkedIn
Vijay Soni
CEO

Business operations, commercial growth and company execution.

Monika Bansal
Executive Director

Company operations, coordination and strategic execution.

Core team

Lovejeet
Python Head

Backend and voice pipeline engineering.

Shubham
AI Engineer

Conversation quality and agent behaviour.

Rinku
TTS & STT Voice Specialist

Speech synthesis and recognition quality.

Uttam
Frontend Developer

Platform interface and the visual builder.

Imran
Head of Digital Marketing

Marketing, demand generation and channel growth.

WHAT WE'RE BUILDING DIFFERENTLY

Decisions we made on purpose.

A visual builder, not one giant prompt

Conversations are built as a workflow you can read, reorder and test — so the person who owns the conversation can review it. See the builder.

India-first multilingual workflows

Built for 32+ languages, including 12 Indian languages and the mixed speech real callers use, rather than treating multilingual support as a later feature.

Browser-first testing

You talk to the agent in your browser before it touches a phone line — no number, no telephony setup, no risk to a customer.

Business context and guardrails

What the agent knows, and what it must never do, are explicit and reviewable — not buried inside a prompt nobody wants to edit.

Reseller and white-label architecture

Partners run client organisations, assign agents and manage credits under their own brand and login domain. Reseller program.

Managed implementation

Support for agent setup, calling workflows and integrations, so a deployment does not depend on the customer having an AI team.

RESPONSIBLE VOICE AI

Voice AI should stay under business control.

An agent that speaks to your customers carries your name. These are the principles the product is built around.

Business-controlled workflows

The agent follows a workflow you designed and can inspect. It is not a black box improvising on your behalf.

Explicit guardrails

Limits on what the agent may say, promise or discuss are configured as rules, and can be reviewed by the people accountable for them.

Human handoff by design

Escalation to a person is part of the workflow. Some conversations should not be automated, and the agent should recognise them.

Appropriate disclosure

Customers should be able to tell they are speaking with an automated agent. We expect deployments to be transparent about it.

Consent and lawful calling

Customers are responsible for consent, disclosure, recording notices, opt-out handling and do-not-call obligations for the campaigns they run.

No unauthorised impersonation

CallMangal is not for impersonating a real person, or for using a voice or identity you are not authorised to use.

Reviewable calls

Recordings, transcripts and outcomes are available in the panel, so what the agent actually said can be checked rather than assumed.

Judgement stays with people

AI output is probabilistic. Transcripts, summaries and extracted fields support decisions; they should not be the only basis for one.

These are product principles, not legal guarantees. See our Terms of Service and Privacy Policy for the terms that govern use of CallMangal.

WHERE WE'VE BEEN

How CallMangal got here.

Foundation

START DESIGNS founded

Began building MVPs and web & mobile products for early-stage founders and growing businesses.

Commerce

EcomHeroes launched

Started designing and developing Shopify stores for ecommerce brands — staying close to real operational bottlenecks.

Product

CallMangal built by START DESIGNS

Started building a white-label AI voice agent tailored for business calling workflows and full brand ownership.

Live calls

From service work to production voice agents

CallMangal moved from client-facing service insight into a product for inbound, outbound and WebRTC voice automation.

What's next

Deeper language coverage and integrations

Continue expanding language support, CRM and messaging integrations, and reliability for high-volume business calling.

CallMangal, In Short

What you're actually getting.

Responsive
designed for natural, low-latency conversation
3 Call Modes
inbound, outbound, and WebRTC
Recorded
call audio and transcripts kept in the panel
32+ Languages
including 12 Indian languages
CRM + WhatsApp
connected actions and follow-ups
EVERY CALL. HANDLED.

Let's put CallMangal on your busiest line.

Book a walkthrough, or hear it on a real call first — no commitment either way.

Partners: reseller program · affiliate program