Ten agents, ten briefs, ten different things they refuse to say.
A voice agent is only as good as the instructions behind it, and generic instructions produce a generic call. Every persona here has its own script, its own vocabulary and, most importantly, its own written list of things it is not allowed to claim.
All ten are live. Pick the one closest to your business and call it. You will learn more in two minutes than from any deck we could send you.
Pick the one closest to your business.
Each page explains where that industry loses leads, what the agent asks, and exactly what it is forbidden from saying.
Real estate
Site-visit booking for new launches and resale. Budget, location, possession timeline and loan status, captured on the first call.
See how it works →Healthcare
Appointment and health-checkup follow-up for hospitals and clinics. Books slots, explains preparation, and never comments on symptoms.
See how it works →Insurance
Term and general insurance follow-up. Advisor tone, unhurried, no closing on the call, and never a premium figure it was not given.
See how it works →Education
Admissions follow-up for institutes and universities. Books counselling sessions and campus visits, and never guarantees a seat or a placement.
See how it works →Automotive
Test-drive booking for dealerships. Brisk, single-purpose, and never quotes an on-road price or an exchange value blind.
See how it works →Hospitality
Room, banquet and event enquiry follow-up for hotels and venues. Captures dates and party size, and never confirms availability it cannot see.
See how it works →Travel
Holiday and itinerary enquiry follow-up. Consultative, understands the trip before pricing it, and never asserts a visa rule.
See how it works →Interior design
Home interiors enquiry follow-up. Scopes the home before pricing, books the site visit, and never invents a per-square-foot rate.
See how it works →Financial services
Investment enquiry follow-up. Fact-finding only, routes to a qualified advisor, and never projects a return or calls anything guaranteed.
See how it works →Fitness
Gym and studio enquiry follow-up. Encouraging, books the trial session, and never promises a number on the scale.
See how it works →Why we build one per industry.
The obvious approach is one agent with the industry supplied as a variable. It works in a demo and falls apart on real calls, for two reasons that have nothing to do with the model's intelligence.
The first is vocabulary. Speech recognition is dramatically better when it knows which words to expect. Kharghar, gynaecologist, MahaRERA, ex-showroom, twin sharing: these are the exact words a general model mangles, and a mangled word in a transcript is a booking error downstream. Each persona ships with its own vocabulary list.
The second is refusals, and it matters more. What an agent must never say is entirely industry-specific. A healthcare agent must not assess a symptom. A financial one must not recommend a scheme. An automotive one must not value an exchange car unseen. Those boundaries cannot be inferred from a variable; they have to be written, reviewed and approved, per industry, before anything goes live.
So there are ten briefs rather than one, and when we build for a new sector, an eleventh gets written with your team rather than assumed.
The rest of the system.
A voice agent is one layer. These are the others.
Before you ask.
My industry is not listed. Can you build one?
Yes, and it is the normal case rather than an exception. These ten exist because we have built them; an eleventh takes the same fourteen days. The work is mostly writing the brief with your team, particularly the list of things the agent must never claim.
Are these demos the same software you deploy?
Yes. Same engine, same interruption handling, same guardrail structure. What differs in a real deployment is that the placeholder blocks are replaced with your actual products, prices and availability.
Can one business use more than one persona?
Yes. A hospital group running both appointment follow-up and health-checkup campaigns would sensibly use different briefs for each, because the questions and the refusals differ.
Which language do they speak?
English and Hindi, including switching mid-call, which is the normal case in Indian sales conversations rather than an edge case. Other languages are scoped at the consult rather than promised in advance.
How much of the script can I change?
All of it. You approve the questions, the objection responses and the refusal list before the agent takes a single call, and you can change them afterwards.
Two minutes on a call beats twenty on a deck.
Pick the persona closest to your business, call it, and try to catch it out.
hello@ensocial.in
Book a consult →