Voicxent · Voice-agent AI platform

Voice agents thatanswer, remind, andfollow up.

For any business that talks to its customers. A Voicxent agent answers the phone, books appointments, reminds people, follows up — and records everything. It does the calling at scale; your people work the queue that needs a human.

Hindi · English · Bengali

Languages spoken

Inbound + Outbound

One platform

Every call, on file

Transcript + recording

What an agent does

One agent.
Every phone job
your team repeats.

Voicxent does not replace your people. It does the calling at scale — and your people work the queue: the escalations, the approvals, and the conversations that need a human.

01

Inbound

Answers inbound calls and books, confirms, or takes a message.

02

Outbound

Makes outbound calls — reminders, confirmations, follow-ups, reactivations.

03

Languages

Speaks Hindi, English and Bengali, and follows the customer if they switch language mid-call.

04

Every call, on file

Records every call: disposition, full transcript, recording and captured fields.

05

One prompt per agent

Dialogue, flow, dispositions and closings in a single versioned prompt. No scattered scripts.

Use cases

One platform, configured per business.

Each use case is a pack configured on the same platform core. The collections pack is live today; other packs are configured for the business that needs them.

Live today

Collections

Verifies the right party, reminds, agrees a promise to pay, settles where allowed, and follows up broken promises.

Configured pack

Clinics

AI receptionist: answers the phone, checks availability, books appointments, sends confirmation SMS.

Configured pack

Garages & service

Books service appointments, reminds customers of bookings, follows up after service.

Configured pack

Pharmacies

Refill reminders and order-ready calls, with confirmation by phone or SMS.

Configured pack

Restaurants

Takes orders and reservations by phone, confirms bookings, reminds guests.

Live today · The collections pack

The pack changes who calls — not the shape of your business.

Same lists in, same files out. In a real portfolio sample (QGBS, Audatec export), this is what the dialling looked like before the pack took over the calling:

76%

of contacts never connected at all

33%

of settlement accounts were the wrong person

23.6%

connect rate — 2,369 attempts on 832 contacts

Five agents, one campaign

Collections pack
AgentWhat it doesOutcomes
RPI VerificationConfirms the right party before any payment talk. No payment discussion until identity is confirmed.8 dispositions
Payment ReminderCalls before the due date. A gentle nudge; intent confirmed; payment date captured.9 dispositions
Promise-to-PayAgrees a specific date and amount. The promise is recorded.12 dispositions
SettlementPrincipal + minimal charges. Settlement letter after approval.11 dispositions
Broken-PromiseFollows up when a promise breaks, and negotiates a new one.11 dispositions

Collections-pack honesty: kept vs broken promises need your payment data — payment truth lives with the bank. The platform works the list you provide.

How it works

From list to results in four moves.

01

Pick a use case

Collections, clinic, garage, pharmacy, restaurant — start with the one that costs you the most dialling.

02

Configure the pack

Agents, scripts, services, language rule — each pack is configured on the same platform core.

03

Run it

Inbound calls answered, outbound campaigns dispatched. Upload leads by CSV and dispatch at once.

04

Get the results back

Dispositions, transcripts, recordings — and an export in the same 16-column shape as your files.

A modern operations team at work while the AI handles the calling

How a call works

  1. 1.Lead or caller in — your list, as-is, or a customer calling you
  2. 2.AI calls or answers — in Hindi, English or Bengali
  3. 3.Identity checked first, where it matters
  4. 4.The right conversation — booking, reminder, promise, or settlement
  5. 5.Everything saved — disposition, transcript, recording, fields
  6. 6.Results back — dashboard and export in your file format

Admin & roles

Four logins. Each sees only its own work.

Run it without calling us: invite callers, organise teams and manage roles from the Users page. The Tenant Admin sets its own voice-provider key — platform values are fallback only. Catalog versions are adopted into your tenant as a copy, never a push.

Super Admin

Role 01

Agent catalog authoring and publishing; tenant creation; Tenant Admin creation.

Cannot see tenant customer data.

Tenant Admin

Role 02

Campaigns; agent adoption; users; leads; settings including provider credentials; exports.

Cannot edit prompts or the catalog.

Team Leader

Role 03

Allocates cases to own team's callers; time-bound settlement approvals.

Own team only.

Caller

Role 04

Own allocated cases only; own disposition updates, audited.

Nothing else. No other team's cases.

Cost rollup — per-tenant cost view for Tenant Admins.Human queue — escalations flagged for people; settlement approvals time-bound and audited.

Security

Your data stays yours.

Multi-tenant, one silo per tenant

One codebase: own database, own subdomain, own secrets. A running campaign stays pinned to its version; new versions never auto-apply.

Secrets and logins

Secrets live in env files with strict permissions — never in code or logs. Logins are JWT-scoped: a user only ever sees their own tenant.

Super Admin is blind

Platform admins cannot see tenant customer data.

Backups per tenant

Own dumps, versioned, per tenant.

HTTPS-only webhooks

Webhook calls over plain HTTP are rejected.

Compliance & audit

Built to be audited.

Consent and calling window

Calls only inside agreed windows.

DNC respected

Do-not-call handling built in.

Every call recorded

Recording + transcript on file for every call.

Number rotation

Caller IDs rotate per policy.

Audited everything

Disposition and lifecycle updates, time-bound approvals.

Letters client-end

Settlement letters issued client-side after Team Leader approval (collections pack).

Said plainly, up front

What Voicxent cannot do.

We would rather you know the edges before you buy than discover them after. These are deliberate boundaries, not missing features on a roadmap.

No live call transfer

Escalations are flagged for human follow-up, not handed over live.

No auto-dialling skip-traced numbers

First call to a skip-traced number is identity check only.

No mid-flight version changes

A running campaign never changes. New versions apply to new campaigns only.

No cross-tenant views

Nobody sees another tenant's customers — by design.

No human dialler in the app

The AI places campaign calls. People work their queue.

Languages

The agent speaks the customer's language.

Hindi, English and Bengali — and it follows the customer if they switch language mid-call. Names are always spoken exactly as written, never translated.

हिन्दीHindi
EnglishEnglish
বাংলাBengali

Who it's for

Built for teams that live on the phone.

Lenders and collections teams, clinics, garages and service businesses, pharmacies, restaurants. If your team spends its days dialling lists where most numbers never connect — or missing calls because nobody is free to answer — Voicxent does the dialling and the first conversation, and hands your people the cases that need them.

How to get started

Step 1

Book a demo

See the agents on real flows for your use case.

Step 2

Pilot on one use case

One agent, one list, measured against your current results.

Step 3

Roll out

Add agents and use cases once the pilot proves itself.

FAQ

Straight answers.

The questions every operations lead asks before a pilot — answered the same way we answer them on a call.

Contact

Book a demo. Run a pilot.

Tell us the use case and your rough monthly call volume. We'll come back with a demo on real flows for your business — then a pilot: one agent, one list, measured against your current results.

Contact us for pricing — no prices are published on this website.