Education Administration /Customer-service agent interactive demo field guide · 8 min

VS365

Handles ~70% of after-hours parent enquiries without a staff escalation.

theater/demos/edu-admin_vs365.html · sandbox · read-only

Open

FIG. 1

The live demo, running on fabricated data. Open it to step through the full flow — every output is shown for a person to approve before anything happens.

How it would work

Reads each inbound student message, classifies what it is and how urgent, drafts a reply grounded in the RTO's own policies and the student's record, and surfaces any welfare or census-deadline signal for a person to approve before anything is sent.

Input 01

The inbound student inbox

Unsorted student messages — fees, special consideration, portal access, placement, campus questions — alongside the student's enrolment record and the RTO's published policies and academic calendar.

Agent 02

Classifies, routes, drafts

Tags intent and urgency, routes each message to the right queue, and drafts a grounded reply citing the policy clause and the relevant census or key date it found. It flags welfare and withdrawal-risk language rather than answering it.

Output 03

A routed draft, working shown

A classified message with a citation-backed draft reply and any welfare signal flagged, laid out for a student-services officer to read, correct, escalate, and approve before a single reply goes back to the student.

Where it works well

It triages and drafts the high-volume routine enquiries every time, and shows the policy clause it leaned on.

At peak — enrolment, census week, results release — the queue is mostly the same dozen question types asked a hundred ways.
Best for a team fielding high volumes of routine enquiries: fees and refunds, special consideration, portal access, key-date questions.
The recaptured hours go back into the students who actually need a person — the distressed, the at-risk, the genuinely complex — not the FAQ traffic.

The slow, invisible cost in a student-services inbox is the sorting — reading every message to work out what it is, how urgent it is, which deadline it touches, and which policy answers it, before anyone can even start a reply.

Where it works badly

It is confidently wrong when its policies or key dates are stale — and a clean, cited reply looks more trustworthy than the data behind it.

Weakest on the messages that matter most — welfare, distress, withdrawal risk — which it should flag for a person, never answer.
If most of your inbox is genuinely complex, one-off, or sensitive, the triage gives you less than your own staff already do.

The honest test

If you cannot say, right now, which version of your fee, refund and special consideration policies these drafts are grounded in, this tool makes your wrong answer faster and better-cited, not safer.

Point it at last semester's fee policy or an out-of-date academic calendar and it drafts a polished, citation-backed reply quoting a census date or refund rule that no longer holds. The citation makes the wrong answer look verified. That is the trap.

What it doesn't do — and shouldn't

It drafts and flags. A person approves and sends. That boundary is deliberate.

WHAT IT DOES

Surfaces the policy clause and the key date each draft reply relied on

Flags welfare, distress and withdrawal-risk language for a human

Shows its classification, urgency and routing for the officer to override

WHAT IT WON’T

Send any reply to a student on its own

Respond to a welfare or distress message itself

Decide a special-consideration, refund or census outcome

A student-services reply can shape whether someone applies for special consideration in time, withdraws before a VET Student Loans census date, or gets help in a crisis. Those carry obligations under the Standards for Registered Training Organisations, the VET Student Loans rules, and a duty of care. A person stays on the decision because the consequence lands on the student and the RTO — not the tool.

What your data has to look like

Current policies and key dates as structured, retrievable text — plus an enrolment record it can match a message to.

Typical readiness

across orgs we see, before the first job

Current fee, refund and special-consideration policies

Usual weak point

An academic calendar with unit census dates

Needs shaping

A routing map of queues and who owns them

Usual weak point

The student's enrolment record, matchable to a message

Needs shaping

A welfare-escalation pathway that already exists

Usual weak point

The real first job

The policies and the academic calendar are usually the weak point — held as PDFs reissued each intake, or dates kept in a spreadsheet someone updates by hand. Getting that into clean, current, retrievable form is usually the real first job — larger and more valuable than the triage layer on top. Once the sources are trustworthy, every drafted reply after that is right by default.

Right fit if…

You field high volumes of routine student enquiries, especially at peak intake and census

The same dozen question types — fees, refunds, special consideration, portal, key dates — repeat constantly

Your fee, refund and special-consideration policies exist as current, citable text

You have a real welfare-escalation pathway a flagged message can route into

Walk away if…

Most of your inbox is genuinely complex, sensitive, or one-off

Your policies live as last intake's PDF that nobody owns or versions

Your census dates and key dates aren't reliably current in any one place

You want a tool that handles welfare and distress messages for you

Open questions

The worried-buyer questions, answered straight

It can draft a wrong answer — which is exactly why nothing is sent on its say-so. Every draft shows the policy clause it cited and the census or key date it used, and welfare or withdrawal-risk language is flagged for a person rather than answered. A student-services officer checks the citation and the date before approving. The tool surfaces the draft; the person stands behind what goes to the student.

It drafts from retrievable text — your fee, refund and special-consideration policies and the academic calendar. If a census date lives in a spreadsheet someone updates by hand, the draft is only as good as that spreadsheet, and the tool can’t tell a stale date from a current one. Getting those sources into clean, current, citable form is usually the first piece of work — and the piece that pays off across every reply after.

No. It removes the sorting and the first-draft typing on routine, repetitive enquiries so officers spend their time where judgement matters — the distressed student, the genuinely complex case, the welfare flag. Every reply is still read, corrected and approved by a person before it sends. The capacity it frees goes back into the students who need a human, not the FAQ traffic.

Current to the active intake. VET Student Loans census dates are set per unit and academic calendars change each intake, so a draft grounded in last semester’s dates will quote a deadline that no longer holds. The honest test: do you know, today, which version of your fee, refund and special-consideration policies and which calendar these drafts are grounded in?

A student’s enrolment record, USI, messages and welfare signals are personal and sensitive information under the Privacy Act and the Australian Privacy Principles. Any deployment runs against your own systems and data handling, not a shared pool — we scope where the data sits and who can see it as part of the build. The demo here runs entirely on fabricated data; Priya Nair is not a real student.

It does not try to answer it. The agent flags welfare, distress and withdrawal-risk language and routes it into your existing wellbeing-escalation pathway for a named person to follow up — the draft beside it is context for that person, never an auto-reply. The duty of care stays with your staff; the tool only makes sure the signal is seen and not buried under routine traffic.

What it takes to build

3–4 weeks · 4 phases

Reused from template~70%

Bespoke to this skin~30%

stack · Claude · retrieval · review UI

What it would cost

Fixed scope, fixed price, fixed dates.

Bite-sized first piece

One contained change, low risk

Pilot build

Most builds land here

Embedded support

Scale on proof

Considering this for your student-services team?

The honest place to start is a bite-sized first piece — one contained change, low risk. Tell us where the inbox hurts; we'll play it back, scope it, and show you what's possible.

Book a call How we work