VANI OS · INVESTMENT BRIEF

This isn't software you buy.
It's revenue you didn't have.

Vani OS turns the inquiries your university or short-course programs are already generating into enrolled students and filled batches — and it only gets paid when that actually happens. This brief is the investment case: the pain by segment, what your stakeholders make, how it works, and exactly what you're being asked to approve.

See the investment ↓ Setup + platform for 4 months: ₹21,00,000 fixed. Everything past that is paid from money the outcome already made.
THE ACTUAL PRODUCT
There's no license fee for the software. Setup and platform keep the lights on — the only line that scales with anything is the outcome fee, and it only ever appears next to an admit that wasn't going to happen anyway. You're not buying a seat count. You're buying enrolled students.
01 · THE PAIN

Two very different problems wearing the same name: "more leads than we can call."

A university applicant and a short-course sign-up don't fail for the same reason. Treat them the same and you fix neither.

University / Full-Degree Programs

12,000 inquiries a cycle. The other 82% just leave.

  • A 20–25 seat calling team, ~75% of dials spent on leads that were never going to close.
  • Decisions take weeks — parents, fee comparison, campus visits — and a cold lead re-enters that decision with a competitor, not with you.
  • One careless sentence on "will I get in?" is a compliance and reputational risk, not just a lost lead.
  • Counsellors call cold — no memory of the last touch, no program-fit context.
Baseline conversion
18%
Admits at baseline
600
Online / Short Courses

30,000 sign-ups. A batch that fills — or doesn't — by Friday.

  • Far higher volume, lower ticket — the economics only work if follow-up is near-instant, not next-business-day.
  • The decision window is hours, not weeks — a webinar attendee decides same-day or is gone for good.
  • WhatsApp is the primary channel here, not a fallback — and most of these leads go cold inside 24 hours.
  • Outsourced BPO floors are measured on dial count, not on whether the seat actually filled.
Baseline conversion
10%
Admits at baseline
600
02 · WHAT YOU MAKE

Not what Vani OS earns. What your institution does.

Same 12,000-lead cycle. Same team, sized down. Three people in the room, three different numbers that matter to them.

Board / Leadership
+₹17.5 Cr

Incremental revenue — 1,000 more enrolled students than your baseline, from the exact same inquiry pool you already have. Nothing new to spend on lead generation.

Admissions Head
50% leaner

A 20–25 seat calling floor becomes 10–12 elite closers hitting the same or a higher target — AI handles triage and follow-up, your people close.

Finance
10.23×

Return on every rupee spent — setup, platform, and every outcome fee included. ₹15.79 Cr net benefit after all of it is paid.

₹48–62Lsaved / year on headcount alone
Real enterprise AI-agent deployments put a fully-loaded counsellor — CTC, infra, benefits, training, attrition — at roughly ₹40,000/month. A 20–25 seat floor costs ₹96L–₹1.2 Cr a year before a single lead converts. Sizing that floor down to the 10–12 elite closers this system is built for saves ₹48–62L annually — on top of, and separate from, the ₹17.5 Cr incremental-revenue number above.
03 · HOW WE MAKE IT HAPPEN

The four things that make this safe to hand a phone to.

Not a chatbot with a script. A counsellor constrained by rules that don't bend, watched by people who can step in.

01

A state machine, not a free-form chat

Every call moves through a fixed sequence — it cannot skip from "what's the fee" straight to "you're in." Whatever the AI says next is shaped by which stage the conversation is in, not by what would sound persuasive in the moment.

OpeningDiscoveryFitRecommendObjectionCommitmentEscalating
02

Guardrails that fire on phrase, not on judgment

Admission decisions, fees, visa/legal, minors, binding commitments — matched deterministically. The same rule fires identically on every call, every time, regardless of how the model is otherwise behaving that day.

LIVE
caller:"will i definitely get admission?"
rule:Category 5 — Admission Decision
bot must never say:you're confirmed
"I can't make admission decisions myself — that's done by our admissions committee after reviewing your full application."
03

A human is always one click away

A supervisor watching the console can listen in silently, barge in on a live call, or take it over entirely — the transcript and event log stay continuous. Escalation doesn't mean the call goes dark.

04

Two dashboards, because two people ask different questions

Operations asks "is the AI actually working, call by call." Finance asks "are we getting outcomes for what we're paying." Neither dashboard substitutes for the other — both recompute the moment a real event lands.

Voice-AI Analytics
"Is the AI actually working?"
Sheets Dashboard
"Are we getting outcomes for what we pay?"

This is the condensed version. Watch it happen to one real lead, end to end →

04 · THE INVESTMENT

What we're actually asking you to sign off on.

One number is a fixed commitment. Everything past it is paid out of money the system already made — never before.

Fixed capital — sign-off needed today₹21,00,000
Setup — one-time₹5,00,000
Platform — 4 months × ₹4,00,000₹16,00,000
Outcome fees — 1,000 incremental admits × ₹15,000
only accrues after an admit above baseline is verified
₹1,50,00,000
Total, if every projected outcome lands₹1,71,00,000
10.23×
Return
On every rupee spent
Net benefit
₹15.79 Cr
Cost per incremental admit
₹17,100
05 · CLOUD OR ON-PREM

Same engine. Different question of who owns what.

This is a data-residency and infrastructure decision, not a different price for the same outcome fee.

RECOMMENDED · SHARED / SAAS

Cloud — managed

Hardware
None to own — pooled GPU capacity, shared across customers
Go-live
Days
Who maintains it
Vani OS, end to end
Bundled into the ₹4L/month platform fee: telephony, compute, and WhatsApp send costs — nothing separate to reconcile. This is why it's a shared pool, not a dedicated box per customer — a dedicated cloud GPU for one institution alone doesn't pencil out for either side.
SOVEREIGN

On-Prem — you own the box

Hardware
~₹4–4.5L one-time (RTX 4090-class tower, sized for a 10–12 seat closer floor)
Go-live
Weeks — hardware lead time + on-site install
Who maintains it
Your IT, Vani OS supports the software layer
You bear directly: the GPU hardware, voice lines (~₹0.60–1.20/min via your own Exotel/Plivo account), and your own WhatsApp Business Account. In return, the platform fee comes down from ₹4L/month — the exact number is scoped once your hardware is quoted, since we're no longer bundling telephony or compute into it.
Sizing it honestly: a single on-prem box, sized to a 10–12 seat closer-floor equivalent, comfortably covers a University-scale cycle (12,000 leads) end to end. We model ~7–9 min per AI call as the real average — our own testing, not the faster 4–6 min a scripted bot might claim — because a genuine multi-turn discovery conversation runs longer than an IVR script, and we'd rather size for that honestly than undersell it. At Online-program scale (30,000+ leads, higher touch-frequency), that's exactly where Cloud's elastic shared pool earns its keep over a single fixed box — which is the concrete reason Cloud is the default recommendation unless data sovereignty is the deciding factor.
06 · WHY IT COSTS WHAT IT COSTS

"AI calling is ₹1/min" is true — and not the whole story.

₹1/min is real for raw telephony minutes alone — the SIM-card cost of dialing, nothing more. A full AI counsellor is a different product entirely.

What's being pricedReal rate
Raw telephony only — carrier minutes, no AI (Exotel/Plivo, India outbound)₹0.60 – 1.80/min
Vani OS's real cost — self-hosted STT/LLM/TTS, telephony + amortized compute only~₹0.60 – 1.20/min
Published enterprise AI-agent platforms — bundled STT + LLM + TTS + hosting₹3.50 – 5.50/min
Premium enterprise AI-agent platforms, same bundle₹13 – 32/min

A full AI counsellor — the one that scores leads, enforces guardrails, hands off compliance-sensitive questions, and reconciles every fee to a ledger your finance team can open directly — runs ₹3.50 to over ₹30 a minute across the platforms we checked, real published and quoted pricing, not estimates. Ours doesn't, because we don't buy that layer from anyone — we built and self-host it. That's what the platform fee actually funds: the infrastructure and engineering that keep our per-call economics below every third-party alternative — not a telephony reseller markup.

You don't pay us to make calls.
You pay us when a seat fills.

Lock the baseline →