All insights

Universities

Why Universities Should Not Buy Chatbots First

· 9 minute read

If the first purchase is a bubble on the homepage, you have bought a press note. Buy a verb that already exists on a desk.

Every August a campus launches a chatbot named after a river or a poet. The press note says 24×7 student support. By September the bot is quoting last year's calendar, leaking a fee balance to anyone who types a roll number, and being screenshotted in a student group for a joke it made about ragging. The computer centre is told to 'tune the personality'. The personality was never the defect. The defect was that the first purchase was a mouth.

This is an opinion for vice-chancellors, syndicates and the deans who are about to sign a chatbot GeM bid because a neighbouring campus did. The opinion is not that conversation is evil. It is that conversation is a surface. Surfaces need something underneath: a ticket store, a fee matcher, a pinned circular, a service account that cannot see marks. Buy those. Put a small mouth on them later if you must.

Prcept builds agents, not mascots. We will lose some early conversations by saying this. We would rather lose those than inherit a poet-bot in year two.

A mouth without a desk is a second queue

Students already have WhatsApp groups, a website, a counter and a rumour. A chatbot that cannot write into a desk's system of record is a fifth channel that the clerk does not read. Deflection metrics will look healthy because students stop trying. That is not support. That is a quieter failure.

The first test of a vendor demo is simple. Where does this ticket live tomorrow if we switch you off. If the answer is 'in our cloud conversation store', you have bought a diary, not an operation.

NAAC and NIRF do not give marks for a mascot. They give work to IQAC. Do not let a chatbot become the quality story. Let a closed geyser ticket become the story.

Conversation multiplies privilege errors

A form posts known fields. A chat invites the student to provide a roll number, a date of birth, a mother's name, a screenshot. The model, being helpful, will look up whatever its service account can see. Helpdesk plus ERP-read is how a stranger learns a fee balance.

Retrieval-augmented generation without a pin will quote a dead circular fluently. Fluency is not a control. The control is a versioned object and a refusal to paraphrase a right.

Jokes, empathy and 'buddy' personas go wrong on ragging, suicide language and caste slurs. A first version should be bland. Bland is a safety feature.

Buy in this order. Invert it and you will fund a redesign with a scandal.
OrderBuyWhy first
1Data map and service accountsSo the mouth cannot see marks
2A real desk system (tickets, matcher, ILS, admissions file)So there is a tomorrow if the vendor dies
3Retrieval over pinned circularsSo facts have versions
4Staff desk cards and a paste banSo humans do not become the leak
5A small conversational surface, logged identity, no personalityNow it has something to say that is true

The press note is the enemy of the first project

A matcher for two programmes will not photograph well. A poet-bot will. Syndicates that want the photograph will pay for the incident. Brief the vice-chancellor: the photograph can be a warden closing a ticket on a phone, if you need a photograph.

GeM categories make this worse. It is easier to buy 'chatbot solution' than 'read-only join of bank scroll and ERP'. Write the latter anyway. If the category does not fit, the file should say why you are not buying the category.

Affiliated colleges will copy the university homepage bot and point it at their own ERP with one credential. Multi-tenant isolation is not a year-two feature. If you must have a surface, have ten isolated mouths, not one brain.

What to tell the syndicate in four sentences

We will not buy a homepage bot this quarter. We will buy a hostel ticket matcher or a fee join. We will put a status line on it after it works. We will keep marks and discipline out of any chat identity.

If they still want the poet, give them a static FAQ that is a pinned PDF. That is a website. Websites are allowed.

How to kill a bot you already bought

Turn off ERP-read today. Leave a prospectus-only retrieval if you must keep the bubble for a week of orientation. Publish a one-line notice: answers on fees and results now require login. Students will complain for three days. That is cheaper than a leak.

Export the conversation store to your file system and revoke the vendor's admin. If you cannot export, you have learned the exit lesson the hard way. Put that sentence in the next purchase file.

Then start the desk-backed pilot. The syndicate will ask why the poet is quieter. Show them the orphan-UTR graph or the geyser SLA. If they still want the photograph, photograph the plumber.

Objections you will hear — and what to do with them

These are the lines that stall the file. Answer them in the room, then put the answer in the note. A spoken answer without paper will be forgotten by the next officer.

Students expect ChatGPT.

Students expect answers that do not change their rights. They already have ChatGPT. They do not need you to become a worse one with their marks behind it.

We will look backward.

You will look careful. After one neighbouring scandal you will look prescient. Careful is a permitted brand.

Our vendor includes the chatbot free with the ERP.

Free is how the service account gets ERP-read. Price the isolation. If they cannot split accounts, the free mouth is expensive.

Conversation is how first-generation students ask.

Then design the later surface for them — stepwise, human door, no jokes — on top of a desk. A mouth with no desk still fails them, only faster.

A quarter without a homepage bot

Announce the pause so deans stop promising the poet in orientation.

  1. Week 1: syndicate pause. Inventory shadow bots already live. Halt the ones with ERP-read.
  2. Week 2–8: run one desk-backed pilot from this cluster — hostel tickets or fee match.
  3. Week 9–10: optional status line with login. No personality. No roll-number trivia.
  4. Week 11–12: write the year-two conversation design, if the pilot was boring in the good way.

How this shows up in the file

Note to the VC: we are not buying a chatbot; we are buying a verb. The press note can wait. The service account will not see marks. If a free ERP bot exists, isolation is a condition of keeping it.

Attach the neighbouring incident, if there is one, without naming and shaming. Syndicates remember stories.

This article is informational field guidance for Indian public universities and colleges, not legal, regulatory, academic, labour or procurement advice. Confirm against the live AICTE / UGC / NAAC notice, the DPDP Act and Rules as commenced, your Act and statutes, reservation orders, labour counsel and the syndicate before you file it.

How this works on an Indian campus

A P3 University should be able to run “Why Universities Should Not Buy Chatbots First” without importing a US playbook. “university chatbot mistake” hits UGC/AICTE/NAAC clocks, exam secrecy, reservation rules, and students who may be minors.

If the first purchase is a bubble on the homepage, you have bought a press note. Buy a verb that already exists on a desk. DPDP applies to student personal data. Chatbots are not a strategy. Exam and admissions writes stay with officers. Affiliated colleges need isolation, not one shared index.

  • No production student data in a vendor SaaS sandbox.
  • Write the academic integrity policy before the tool.
  • Consent and purpose tags on student-facing agents.
  • Budget for staff training, not only licences.

Close this loop before the next CAB

Put “Why Universities Should Not Buy Chatbots First” on the next change-advisory or bid-opening agenda as a single line item with an owner. If it cannot earn a line item, it will not earn a control. The owner should be a P3 University, not “the vendor.”

Revisit the item when the model, the GeM term, the region, or the SI changes. “university chatbot mistake” is not a one-time workshop. It is a watch item. Date the last check. Unsigned watch items are souvenirs.

What must be true before you file this

If “Why Universities Should Not Buy Chatbots First” is only a heading, it will not survive a file inspection. A P3 University should be able to attach one artefact that proves “university chatbot mistake”: a log export, a clause, a scored row, a dated notice, or a refusal rule.

Write three dated sentences: what was decided, who owns it, and when it will be re-checked. Unsigned sentences are souvenirs. Dated sentences are controls.

  • Name the owner of “university chatbot mistake” inside the institution.
  • Attach one artefact a stranger can open next year.
  • Revisit when the model, the notice, or the SI changes.
  • Do not treat a vendor slide as evidence.

Questions this usually raises

Are all campus chatbots a mistake?
A conversational surface on a real desk, with login and a pin, can be fine. A homepage personality with ERP-read is the mistake. Buy in that order.
What should we buy first instead?
Usually hostel tickets, fee reconciliation, or circular retrieval. Something a clerk already understands and an auditor can sample.
Does NAAC expect a chatbot?
We have not found a NAAC requirement to deploy a student chatbot. Do not invent one to justify a purchase. IQAC has enough real work.
Can we keep a bot for the prospectus only?
Yes, if it retrieves the live prospectus and cannot see student records. Treat it as a website with a query box, not as an ERP front end.

Sources