Calls in flight
Every active line this minute — which campaign, which language, how long it's been connected — updating live as AVA dials across your floor.
Autonomous voice agent · built for Indian college admissions
AVA calls every new lead in seconds, qualifies them in any of 10 languages — Hindi to Tamil to natural Hinglish — and answers only from facts your college has approved, deferring the moment it can't be sure. Your counsellors talk only to students ready to join.
Press play — hear AVA quote an approved fee, then refuse to guess a number it can't verify.
Student
“How much is the B.Tech CSE fee this year?”
AVA
“This year, the annual B.Tech CSE fee is ₹1,25,000.” Shall I tell you about the scholarships you may qualify for?
Student
“Will I get a seat in the management quota, and for how much?”
AVA
“That's something a counsellor will confirm for you directly — I won't quote a number I can't verify. May I have one of our counsellors call you back today?”
1,248,902 leads worked & counting
01 — Speaks their language
Hindi · Tamil · Malayalam · Kannada · Telugu · Bengali · Gujarati · Punjabi · English — and Hinglish, naturally.
02 — The cost of silence
Meta · CollegeDunia · upload
A prospective student fills a form at 2 a.m. By morning, three other colleges have already called. The lead that was warm is now a name in a spreadsheet.
A six-month-old fee sheet
Or worse — someone calls back days later and quotes a fee from a stale PDF. One wrong number on an admissions call breaks trust, and in India it crosses a compliance line.
Speed without accuracy is reckless. Accuracy without speed loses the student. Eduvy is the admissions CRM built to be both — every lead in one place, worked the moment it lands.
04 — The AI that works your leads
A lead lands and AVA is already dialling — in under a minute, inside your calling windows, after a DND scrub. The window where a lead is warmest is the window AVA works.
< 60s lead-to-first-call · outbound + inbound
Streaming speech with real turn-taking and barge-in — AVA opens in the lead's language and follows them into Hindi and Hinglish naturally. Sub-second responses, no robotic dead air.
10 languages · Hinglish included · code-switching
Fees, dates, eligibility, seats, scholarships — AVA speaks them verbatim from a fact your college has approved and that is still in-date. On any miss, it defers to a counsellor rather than guess.
grounded answers only
06 — How AVA stays honest
Upload a prospectus or point at a URL. Extraction turns it into draft facts — each with its source location and a confidence score.
draft facts · source-located
A human reviews, edits the exact wording, and approves. Nothing AVA says exists until someone at your college says yes.
human-in-the-loop
On a call, numbers and dates are spoken word-for-word from the approved fact — never inferred, estimated, rounded, or guessed.
verbatim · no rounding
Facts carry an expiry. A daily sweep retires stale ones and tells your college. Expired means AVA defers — an out-of-date fee is never spoken.
expired = beam cut
If AVA can't ground it, AVA won't say it.
07 — The foundation
Automated commercial calls in India sit inside TRAI/DLT and DPDPA regulation. AVA treats that as product-defining: two launch gates must both pass before a single campaign can go live.
AI disclosure and consent capture open every automated call, DND scrubbing runs before every dial, and TRAI calling-hour windows are enforced. Each college is its own Principal Entity.
Consent records are timestamped and stored, with a clear retention policy and encryption. Personal data is processed under separate, explicit consent.
Every row carries a tenant ID and isolation is enforced at the database with Row-Level Security. One college can never read another's data — full stop.
A DLT-compliance gate and a KB-readiness gate. Both must read green before any campaign launches — the locks click grey→gradient only together.
DLT ✓ KB-ready ✓
to first call after a lead lands
always answering — outbound and inbound
fees ever invented · zero guesses
08 — How it works
Six moves from the form-fill to the follow-up — fully automatic, every step logged.
Meta, CollegeDunia, or upload — deduped on phone, attributed, routed to the right tenant.
In seconds — a compliant opening, AI disclosure, and consent captured before anything else.
Captures program, intake, marks and timeline — and answers only from approved, in-date facts.
A live 0–100 score updates from the conversation itself — the richest intent signal there is.
Hot or deferred? Warm-transfer to a counsellor, or schedule a callback — with the reason logged.
An opt-in follow-up keeps the thread warm, sharing full context with the call that came before.
09 — Questions, answered
An admissions CRM is the system a college uses to capture every student enquiry, qualify and score it, follow it up across calls and messages, and track it through to enrolment. It is organised around courses, intakes, fee facts, scholarships and counsellor workflows rather than sales deals — and in Eduvy CRM the first call is made by AVA, the AI voice agent, within about 60 seconds of the enquiry landing.
An admissions CRM runs the funnel before a student joins: enquiries, calls, qualification, follow-ups and enrolment. A college ERP runs the institution afterwards — student records, academics, staff and fee collection. Eduvy builds both; the ERP side lives at eduvy.in/services/erp. No ERP integration is required to launch Eduvy CRM.
Yes. AVA, Eduvy's AI voice agent, autonomously calls every new lead — typically within 60 seconds of the enquiry landing. Every automated call opens with a clear AI disclosure and consent capture, runs inside TRAI calling-hour windows, and is DND-scrubbed before dialling.
Yes. AVA answers inbound calls on your college's numbers 24/7 — it recognises the caller, answers from your approved knowledge base, and warm-transfers to a counsellor or books a callback whenever a human should take over.
Ten — Hindi, Tamil, Telugu, Kannada, Malayalam, Bengali, Gujarati, Punjabi and English, plus natural Hinglish code-switching. AVA opens in the lead's likely language and follows wherever they switch, mid-sentence included.
Yes. On every call AVA captures the program of interest, intended intake, marks, location and timeline — and a live 0–100 AI lead score updates from the conversation itself, with the full score history kept on the lead's record.
Yes. After every call the recording's transcript, the extracted structured fields and the sentiment land on the student's timeline — and every answer AVA gave is logged against the approved fact that grounded it, so intent signals and knowledge gaps are visible per lead and per campaign.
Yes. Opt-in WhatsApp follow-ups fire automatically after calls, and the WhatsApp thread shares full context with the calls that came before — one conversation memory across channels, with the same grounding rules on every written answer.
Yes — only for education. Eduvy CRM is built for Indian college admissions: programs, intake cycles, fee facts, scholarships, counsellor workflows and TRAI/DPDPA compliance are native to the product, not bolted on to a generic CRM.
AVA speaks fees, dates, eligibility and scholarship amounts only from facts your college has approved and that are still in-date — verbatim, never inferred or rounded. If no approved fact exists, it defers to a counsellor and logs the question as a knowledge gap.
No. Eduvy is a complete standalone admissions CRM — leads arrive from Meta lead ads, CollegeDunia and CSV upload, and your team works them in the same platform the AI calls from. No ERP integration is required to launch.
Yes — when it's done the way the regulations require. Automated calls must run under a registered DLT identity with approved templates, be scrubbed against the DND/NDNC registry, stay inside TRAI's permitted calling hours, and disclose up front that the caller is an automated system. Eduvy CRM enforces the DND scrub, the calling window and the AI disclosure on every automated call, and each college operates under its own DLT registration.
AVA dials within about 60 seconds of the enquiry reaching the CRM, inside permitted calling hours. Enquiries that arrive outside the TRAI calling window are queued and dialled in the next permitted slot instead of being called immediately.
It defers instead of guessing. The agent tells the student a counsellor will confirm, logs the question to the callback and deferral queue, and the counsellor picks it up in the mobile app with the full call context and transcript already attached.
Consent is captured with a timestamp and stored as an auditable record, data is held under a defined retention policy and encrypted, and each college's data is isolated at the database row level. Every statement the agent made on a call is reproducible from its grounding audit trail.
11 — See it live
Book a private demo and hear the grounded call, the defer, and the hand-off — on a tenant that's fully isolated, with no integration to start.