A counsellor agent that replies to every lead on WhatsApp within a minute, qualifies and books the demo class, and keeps the CRM honest. A doubt assistant that answers from your modules and routes what it cannot to faculty. Faculty keep the teaching, counsellors keep the close.
Thirty minutes, free. You leave knowing which workflow to start with, what a pilot covers and what it costs.
PilotAI Agent Pilot: one workflow, 2–3 weeks, $4,500 (₹3,50,000) fixed. What the pilot includes.
Who this is for: Test-prep institutes from one centre to multi-city chains, and online programmes, where counsellors cannot call fifty fresh leads in the first hour and faculty cannot answer every doubt at 11 pm.
A parent fills in a form at 10 pm. Your counsellor calls at 11 the next morning. Two competitors messaged within ten minutes. Results day and new-batch season make it worse, exactly when it matters most.
Counsellors update the CRM at the end of the day from memory. Readiness, exam, class and preferred batch are missing, so the follow-up sequence is the same message to everyone.
Students message faculty on WhatsApp at night. A general chatbot answers with a confident wrong method or a different approach from the one your faculty teach. Faculty answer the same rotational mechanics question thirty times.
Trigger: A parent submits the Meta lead form at 10:14 pm: "Class 11, JEE, Nagpur".
| Measure | Today | With the agent |
|---|---|---|
| First response to a new enquiry | Next working day | Under a minute, including nights and Sundays |
| Counsellor hours on first contact and qualification, per month | ~40 hours | ~6 hours (exceptions and callbacks) |
| Leads with exam, class, batch and readiness recorded | Whatever got typed in at 6 pm | Every one |
| Demo-class offer made | When the counsellor reaches them | In the first conversation |
Worked example. The assumptions:
Counsellor time is the smaller number. The one that matters is how many of the 300 attend a demo class and enrol, which is why a pilot measures demo bookings and enrolments against the previous intake rather than quoting an uplift here.
Put your own numbers in the automation savings calculator, or get a price range from the AI agent cost estimator.
Each agent has a small set of tools it is allowed to call and nothing else. That list is the guardrail, and it is written down here so you can see it.
On WhatsApp, in the family’s preferred language. Qualifies, answers from approved material, books the demo or counselling slot, writes a summary and readiness score to the CRM, and runs the approved follow-up sequence. Hands over the moment fees are negotiated or the family asks for a person.
Answers student doubts from your modules, lecture notes and solved examples, cites where the answer came from ("Module 7, Example 4"), gives hints rather than full answers before a test, and routes anything outside the material to a faculty queue grouped by topic.
Watches attendance, test scores and submissions. Nudges the student with something specific ("you missed two chemistry classes; recordings are here, doubt session Thursday") and alerts the batch mentor. Never messages a parent without a mentor reviewing it.
We built Talking Books, an AI learning platform whose tutor decides what to explain next from live student context rather than following a fixed script, with adaptive exercises and mastery tracking across eight question types. The counsellor agent is a reference build you can try below. Institute results will be published when an institute agrees to publish them, and not before.
It can help accurately when it is grounded in your material and explicit about sources. A general chatbot solving a physics problem from memory is a liability. Your assistant retrieves from your modules, cites them, is Socratic before tests, and routes what it cannot answer to faculty. Faculty audit what it said.
It follows a sequence you approve, with caps on frequency, stops the moment anyone asks, and never messages about a student’s performance without a mentor reviewing the message. WhatsApp’s own rules on templates and opt-in apply and we design to them.
Yes. It replies in the language the family writes in, from the same approved material. For scripts and regional languages we test the output with your counsellors before go-live, because a confident mistranslation of a fee structure is worse than English.
The AI Agent Pilot covers the counsellor agent on WhatsApp for one centre or one programme: fixed fee, two to three weeks, measured against your last intake. The doubt assistant and at-risk agent are scoped builds after that. The cost estimator gives you a range for your setup.
CRM automation sends the same message to everyone on a timer. An agent reads the reply, answers the question that was actually asked, books a slot and writes down what it learned. Keep the CRM; the agent makes what is in it true.
Which exams, how many enquiries a month, which CRM. We reply with what a pilot on your admissions flow would cover.
Or start with a number: the AI agent readiness quiz takes three minutes and gives you a score and a 30-day plan.