When people say "AI in the ERP" they usually mean a chatbot bolted to the side and a couple of dashboards with the word predictive on them. The interesting opportunity is much bigger and almost entirely unrealised: a school ERP has been collecting the operational record of the school for years, and that record is the most underrated AI substrate in the building.
This post is about what AI on top of a complete ERP actually unlocks, beyond the surface-level features.
The data you already have, sitting there
An Education ERP that's been live for three years has accumulated, without anyone thinking of it as "training data":
- Every admission — enquiry to enrolment funnel, with seasonal patterns and conversion rates per source.
- Every attendance record, by class, by student, by day.
- Every fee transaction — paid, partial, defaulted, refunded — with timing.
- Every academic record — formative and summative scores, term by term.
- Every parent communication — broadcast, complaint, follow-up.
- Every staff event — hired, transferred, left, with reasons.
This is a structured longitudinal record of how the school actually operates. It's not magic, and it's not "training data" in the LLM sense. It's the raw material for the kind of pattern-finding that's been impossible to do manually.
What AI on top of this actually enables
1. Admissions, treated as a real funnel
Most schools run admissions on tradition: a calendar, a few ads, an open house. The ERP has data on which enquiry channels actually convert, which weeks of the year are the most efficient, which counsellors close at what rate, and which student profiles cluster at which fee bands.
An AI layer over this isn't predictive in any fancy sense — it's a competent admissions analyst on call: "enquiries from this source convert at 28%, the others at 11%. Spend accordingly."
2. Fee defaults seen before they happen
Fee default has a shape. Late by 10 days, then again next month, then a partial payment, then silence. The ERP has the full timing record. An AI looking at it can flag the families heading toward default a month earlier — not to send a stern reminder, but to have a quiet finance-office conversation about a flexible plan before the relationship sours.
The win isn't predicting defaults. It's predicting them early enough that a phone call still solves them.
3. Attendance patterns that mean something
A student dropping from 95% to 85% attendance over two months is the kind of thing a class teacher might notice. A student dropping at the same time their formative scores dip, their late-fee-payment days creep up, and they stop attending PT — that combination is something only a system looking across modules can spot.
This is the AI use-case that schools should care about most: the early intervention signal. Not a predictive model with a black-box score; a transparent rule-set that surfaces students whose pattern across modules has shifted, so a counsellor can step in.
4. Communication that doesn't sound like a robot
Parent communication at most schools is one-to-many SMS blasts. The ERP has every parent's child, class, subjects, attendance, recent results, and fee status. AI-drafted parent communications can be substantive — "your daughter has been doing strong work in maths this term and her teacher mentioned she's been mentoring others" — rather than generic.
The teacher signs off; the AI does the drafting. The newsletter that used to take an hour takes twelve minutes.
5. The principal's morning brief
The most useful AI feature in an Education ERP isn't a chatbot. It's a one-page morning brief — automatically generated, in plain English — that surfaces what changed since yesterday: enrolment movements, attendance anomalies, fee collections, exam reviews pending, parent escalations to address.
Nothing predictive. Nothing magical. Just the operational pulse of the school summarised, every day, by 7 AM.
What this isn't
This isn't "we added a chatbot to the ERP." A chatbot in an ERP is mostly a worse search box. The opportunity is in removing the human's manual data-stitching step — letting AI do the cross-module joins that nobody has time to do, and presenting the result as something a school leader can act on this morning.
It also isn't AI-doing-the-decision. The principal still makes the decision. The counsellor still has the conversation. The finance head still works out the plan. What changes is that they get the right signal at the right time, instead of finding out at the end of the term.
The honest sequencing
If you're a school considering this, the honest sequencing is:
- Get a real ERP collecting clean data across modules for at least a year. You cannot AI your way out of a fragmented data layer.
- Start with the descriptive AI features — daily brief, cross-module anomaly detection, communication drafts.
- Add the early-signal layer next — fee default risk, attendance drift, academic slide.
- Only then experiment with the predictive layer, with full awareness that it's the most error-prone.
The ERP is the substrate. The AI is the second layer. Without the first, the second has nothing real to work on.
Send us a term-end paper. We'll grade it free.
No commitment. We'll upload it to Skora AI, run it end-to-end, and email you back a graded sheet with per-question audit trail, weak-area analysis, and our price for a pilot. Most schools get the grade back the same day.