AdminAdvisor · Module 1 — AI Admissions Advisor · Complete solution architecture

The complete blueprint
for IMAS's AI
admissions advisor

This document explains the full system — what it does, how it works, how it will be built, and who does what. Written so that anyone on the team — technical or not — understands the same thing.

6mo
Avg parent decision window
21
KB sections covering IMAS
3
Parent channels at launch
90%
KB fill before AI goes live
What this system replaces
2–3 staff chasing parents via Social media
50-field admission forms nobody completes
3-day response times to enquiries
Parents lost because nobody replied at 11pm
No visibility into where parents drop off
What it delivers
AI advisor available 24/7 in EN, BM, and Arabic
Contact captured through trust, not gates
3-second response time, always
Application pre-filled from conversation history
Full parent timeline visible to staff
The core idea

What is this system, in plain language?

IMAS's AI admissions advisor is a system that knows everything about the school — fees, curriculum, boarding life, visa process, culture, and more — and talks to parents in a natural conversation across Social media, the school website, and the reception tablet. It never sleeps, never gets frustrated, never forgets what a parent said three months ago, and never gives a wrong fee figure.

The school with the fastest, most helpful, most personalised first response wins. Right now, that school probably is not IMAS. This system fixes that — permanently.

Root problem this system solves — parents lost early because the school cannot respond fast enough, well enough, at 11pm on a Sunday.
01
The parent talks to an AI advisor

Via Social media, the school website, or the reception tablet. The AI knows IMAS completely — in English, BM, and Arabic. It answers any question instantly, remembers the parent across months, and never asks them to repeat themselves.

No forms to fill
24/7 availability
Multilingual by default
Session memory across months
02
The system qualifies and nurtures the lead

Behind every conversation, the system scores the parent's intent, detects their family type (GCC, local, expat), and decides when to suggest a school tour or start an application. Staff see a live pipeline with AI-recommended next actions.

Intent scoring 0–100
GCC / local / expat persona detection
Tour and application CTA timing
Staff action queue
03
The AI never says anything wrong

Every fee figure, every accreditation claim, every policy statement the AI says is checked against the school's verified knowledge base before it reaches the parent. If the AI is not certain, it says so and connects the parent to a human.

Fact lock on all numbers
Accreditation verification
Audit log of every response
Safe fallback to human
The parent journey

How a parent goes from stranger to enrolled — in 6 months

Parents do not decide in a day. The average family takes 6 months from first enquiry to enrollment. Today, IMAS loses most parents early — because nobody responds fast enough. This system changes every stage of that journey.

Month 1
Browses fees page anonymously. No reply for 3 days.
Lost here — today
Month 2
Social media query. Generic PDF sent. No follow-up.
Tepid response
Month 3
Attended open day. Gets brochure. No follow-up.
Missed signal
Month 4
Comparing schools. Competitor follows up first.
Competitor wins
Month 5 — with AdminAdvisor
AI follows up with tour invite. Tour booked automatically.
Re-engaged ✓
Month 6 — enrolled
Form 80% pre-filled. Payment confirmed.
Enrolled ✓

How we earn the parent's contact — the trust ladder

Parents hesitate to share their number. We never demand it. We earn it through three tiers — each one feeling natural and fair to the parent.

Tier 1

Anonymous exploration
No contact collected
AI answers everything freely — no gates, no forms
Parent asks any question — fees, curriculum, boarding life, Arabic support, safety protocols, university destinations. The AI answers completely and generously. No registration required. No "fill this form first." The system picks up intent signals quietly in the background and builds an anonymous profile on the device.
"How much is boarding for Form 1?" → Full fee table, what's included, what's not, payment schedule — answered instantly at 11pm on a Sunday.
Auto language detectionNo login requiredIntent signals captured silentlySession memory by device

Tier 2

Soft contact exchange
WhatsApp number — one field
Value first, then the ask — a fair exchange
When the parent requests something with delivery value — a fee schedule PDF, virtual tour link, programme comparison, open day registration — the AI offers to send it directly: "I'll send that to your WhatsApp now — what number should I use?" The content is sent regardless. But ~70% of parents give the number because the exchange feels proportional. Everything they asked before is now attached to their profile.
"Can you send me the fee breakdown?" → "Of course — shall I send it to your WhatsApp?" → Parent gives number → PDF sent → Anonymous history merges to named lead → Intent score updates.
One field onlyValue trade, not a gateFull history preservedDeclining is fine — AI notes intent

Tier 3

Named relationship
Full profile — natural moment
Tour booking or application — contact is expected and welcome
When a parent books a school tour or says they want to apply, giving their name and child's details is completely natural. The system merges their full conversation history — every question across every session — onto the named profile. Staff see a complete relationship from day one. The application form is 70–80% pre-filled before the parent sees it.
"I'd like to book a tour for my daughter" → AI books slot, sends confirmation, creates named profile, pre-fills application with everything already known from months of conversation.
Full history mergedApplication pre-filledTour confirmation automatedStaff see complete timeline from day 1
The system map

How all the pieces fit together

The entire system is one connected platform. There are no separate tools, no switching between apps, no data falling through the cracks. Everything runs from a single parent profile that follows them from first anonymous question to enrolled student.

Layer 1
Where parents talk to IMAS
3 channels at launch
Website
Chat widget on IMAS website
A floating chat panel on the school website. Works on mobile and desktop. Parents can chat anonymously from any page.
WhatsApp / Social Media
WhatsApp Business AI
The school's WhatsApp number is powered by the AI. Parents message it exactly as they would a human — the AI responds in seconds.
Walk-in
Reception tablet kiosk
A tablet at the reception desk. Walk-in parents interact with the AI directly — or staff can enter their details on their behalf.
Layer 2
The channel gateway — normalises everything into one format
Invisible to parents
Receive
Message normaliser
Strips away channel differences. Social media message, website chat, and tablet input all become the same thing internally.
Detect
Language detector
Automatically detects English, BM, or Arabic from the first message — including mixed-language messages common in Malaysia.
Identify
Identity layer
Recognises returning parents by device fingerprint (Tier 1) or WhatsApp number (Tier 2). Loads their full history before the AI responds.
Protect
Consent & compliance
PDPA Malaysia and GDPR consent flows run automatically. Data retention and deletion rights are handled by the platform.
Layer 3
The AI conversation engine — the brain of the system
Core product
Classify
Question classifier
Decides: is this a fees question (use exact number), a process question (use facts + language), or a fit question (use AI reasoning)?
Understand
Persona detector
Identifies family type — GCC relocating, local Malaysian, Western expat — and adjusts what the AI emphasises in its response.
Retrieve
Knowledge retrieval
Fetches the exact right information from IMAS's 21-section knowledge base. Structured facts + narrative understanding, combined.
Generate
Response generator
Creates a warm, accurate, school-specific response in the parent's language. Three paths: exact template, hybrid, or full AI reasoning.
Validate
Fact lock
Every number and policy statement is verified against the source database before delivery. Wrong answer blocked, safe fallback sent.
Layer 4
The knowledge base — everything IMAS knows, structured for AI
21 sections
Identity knowledge
Sections 1–2
Who IMAS is, what it stands for, the brand promise, differentiators, family profiles, admissions journey, CTA rules.
Operational knowledge
Sections 3–18
Placement logic, fees, boarding, pastoral care, English support, visa, transport, policies, SEN — every operational detail.
Intelligence knowledge
Sections 19–21
Conversational FAQ intelligence, objection handling, post-admission journey, and governance rules for AI quality control.
Layer 5
The data layer — where everything is stored securely
Supabase · PDPA + GDPR
Parent profiles
Every parent. Every session.
Anonymous from Tier 1. Every question, every topic, every intent signal — stored and accessible to staff.
Conversation history
Every message, every session
Full timeline from first anonymous question. Never lost between sessions. Parents never repeat themselves.
Knowledge embeddings
Narrative search index
The AI's understanding of IMAS's narrative content — philosophy, culture, boarding life — stored as mathematical vectors for semantic search.
Audit logs
Every AI response verified
Every response the AI generates is logged — what was said, what knowledge it used, whether the fact lock passed or blocked it.
Layer 6
What the staff see — one dashboard, complete picture
Staff interface
Admissions team
Pipeline board
Kanban view of all leads — Anonymous → Soft contact → Named → Applying → Enrolled. Each with intent score and AI-recommended next action.
Admissions team
Parent timeline
Every parent's full relationship history — every message, every topic, every document, every touchpoint. AI-generated summary at top.
Principal
Analytics dashboard
Conversion funnel, seat occupancy by programme, drop-off points, campaign performance, AI insights on what is working and what is not.
AdminAdvisor team
KB review & AI monitor
Reviews KB submissions from school, approves sections, monitors AI response quality, reviews fact lock violations, triggers KB updates.
The knowledge base

21 sections — the complete picture of IMAS

The knowledge base is not a folder of documents. It is a structured, verified store of everything IMAS knows — organised so that the AI can retrieve exactly the right information for exactly the right question. The school fills it in. Our team reviews and approves it. The AI uses it.

⚠
The 90% rule: The AI advisor will not go live until at least 90% of the knowledge base is filled in and approved by our team. Below 90%, the system runs in "preview mode" — parents see a notice that information may be incomplete. This protects IMAS from the AI giving wrong or incomplete answers to real parents.
Part A · Sections 1–2
Institutional identity
Who IMAS is. Vision, mission, philosophy. Accreditation and exam boards. Curriculum identity. Campus snapshot. GCC and family personas. Differentiators. Cultural and religious positioning. Leadership. Brand promise and advisor persona.
Critical — required before anything goes live
Section 3
Student placement logic
Age-to-grade mapping. Curriculum equivalency (Cambridge ↔ Malaysian ↔ American). Transfer student placement. Mid-year entry rules. English proficiency placement levels. Grade advancement policy.
Advisory layer
Section 4
Programme overview
Full programme ladder from preschool to A-Level. Each stage's curriculum, Islamic integration, language structure. University preparation. Co-curricular framework. How programmes differ from each other.
Fact-locked
Section 5
Programme fit scoring
The AI advisor's intelligence layer. Student profile inputs. Parent goal inputs. Fit scoring matrix — which student profile fits which programme best. English readiness thresholds. GCC transition logic. Best-fit and not-ideal case examples.
Critical — advisory AI intelligence
Section 6
Academic standards
Teaching methodology. Assessment and reporting framework. Homework expectations by year group. Academic intervention for struggling students. High achiever development. Teacher qualification standards. Academic performance data.
Fact-locked
Section 7
Student life & daily experience
School day schedule. What a typical lesson looks like. How students make friends. New student onboarding. Islamic daily life experience (Quran, Solat, Tarbiyah). How parents stay connected to their child's day.
Advisory layer
Section 8
Pastoral care & safety
Child safeguarding framework. Physical campus security. Anti-bullying policy and resolution process. Counselling support. What happens when a student is unwell. Emergency response. Escalation triggers — always to human.
Critical — high sensitivity
Section 9
English support & language
English entry requirements by year group. Assessment and placement testing. ESL/EAL bridge programme. GCC and Arabic-background student transition logic. Parent reassurance framework for weak English students.
Advisory layer — GCC critical
Section 10
Fees, payment & financial
Tuition fee by year group. One-time fees. Recurring costs (books, uniform, bus, meals). Payment schedule. Accepted methods. Full cost transparency. Value justification framework. Financial objection handling.
Critical — fact-locked, all numbers verified
Section 11
Scholarships & financial aid
All scholarship types, eligibility criteria, application process, renewal requirements. Affordability pathways and installment options. Common misconceptions the AI corrects.
Fact-locked
Section 12
Visa, relocation & international
Student visa pathway and EMGS process. Guardian/parent stay options. Arrival and settlement support. Housing guidance near campus. GCC family relocation planning framework. High-risk immigration cases always escalated.
Advisory layer — GCC critical
Section 13
Transport & access
Bus route coverage zones. Transport safety and supervision. Monthly cost by zone. Car line pick-up/drop-off system. Honest guidance on Malaysian traffic and commute realities from GCC-preferred residential areas.
Fact-locked
Section 14
Parent communication
Official communication channels and their purpose. Academic reporting schedule. Emergency communication protocol. Daily visibility into student life. Language accessibility for Arabic-speaking GCC parents. Community engagement.
Advisory layer
Section 15
Outcomes & alumni
IGCSE performance indicators. Actual university destinations by country. Graduation qualification pathways. Character and non-academic outcomes. Alumni success narratives the AI uses for hesitant parents.
Advisory layer
Section 16
Culture, values & identity
School archetype. Competitive comparison logic — how IMAS honestly compares to other schools. Child personality fit. Parent identity resonance. Misalignment risks — when the AI is honest that IMAS may not be the right fit.
Advisory layer — sensitive
Section 17
Policies & operational rules
Attendance and punctuality. Uniform and dress code. Behaviour and discipline. Academic integrity. Technology and device policy. Leave and travel rules. Refund and withdrawal policy.
Fact-locked
Section 18
Special cases & SEN
Learning support philosophy. Supported and unsupported student profiles. ADHD and behavioural complexity rules. Emotional and transitioning students. Medical conditions. Custody and legal family situations — always escalated to human.
Critical — always human-escalated
Section 19
Conversational FAQ intelligence
Parent intent detection. Persona signals. Emotional layer recognition. Core response formula. Objection handling library — "fees are too high", "my child's English is weak", "I'm comparing schools." AI tone governance.
Advisory — AI intelligence layer
Section 20
Post-admission & onboarding
Offer acceptance pathway. Payment completion journey. Enrollment documentation checklist. Uniform and books setup. Parent systems activation. Orientation and first-day readiness. International arrival and settlement. First 90-day success framework.
Advisory layer
Section 21
Knowledge governance
Section ownership. Source-of-truth architecture. Version control system. Time-sensitive data governance (fees updated annually). AI hallucination prevention rules. Confidence scoring. Audit and quality assurance. Incident response.
Critical — AI safety and accuracy
The AI engine

How the AI decides what to say — and how it stays accurate

The engine is the most important part of the system. It is what makes the AI sound like a genuine IMAS admissions advisor rather than a generic FAQ bot. Here is how it works — in non-technical terms.

🔍
Step 1 — Understanding the question
Every message from a parent is classified before the AI does anything else. "How much is boarding?" is a fee question — it has a definitive answer that must not be approximated. "Is boarding right for my shy 13-year-old?" is a fit question — it requires empathy, reasoning, and school-specific knowledge. The classifier routes each question to the right response path.
Fee question → use exact RM figure from database. Fit question → use knowledge base + AI reasoning. Policy question → use verified facts + natural language.
👤
Step 2 — Knowing who the parent is
Before responding, the engine identifies what type of family this is. A GCC family relocating from Saudi Arabia has different priorities than a local Malaysian family — the GCC family needs Arabic support, safety, and Islamic environment details most. The engine detects this from signals in the conversation and adjusts what it emphasises in its response.
GCC signals: mentions Dubai, asks about Arabic, asks about visa → emphasise Arabic programme, Islamic environment, safety, EMGS process.
📚
Step 3 — Fetching the right knowledge
The engine does not search all 21 sections for every question. It identifies which sections are relevant — fees for a fee question, pastoral care and boarding for a safety question — and retrieves only those. It combines exact structured data (the fee amount) with narrative understanding (what boarding life actually feels like at IMAS) to give a complete answer.
Parent asks about boarding safety → retrieves Section 8 (pastoral care), Section 7 (student life), and any GCC boarding family testimonials from Section 1.9.
✍️
Step 4 — Generating the response
The response is generated in one of three ways. For fees and exact facts: the database value is rendered in the parent's language by the AI — the number cannot change. For operational questions: structured facts are combined with the AI's language ability. For complex fit questions: the AI reasons fully but is strictly limited to only what is in the knowledge base — it cannot invent details about IMAS.
The AI is never allowed to say "approximately RM 18,000" or "I think the fees are around..." — it says "RM 18,000 per year" or redirects to the admissions team for clarification.
🔒
Step 5 — The fact lock check
Before any response reaches the parent, it is validated against the source data. Every number in the response must match a verified record in the database. Approximation language near fee amounts is blocked. Unverified accreditation claims are blocked. If the response fails the check, it is replaced with a safe message that connects the parent to the admissions team directly.
Every blocked response is logged in the audit table. The AdminAdvisor team reviews these weekly to identify knowledge base gaps and improve AI accuracy over time.
📋
Step 6 — Updating the parent profile
After every response, the system updates the parent's profile. Intent score is recalculated. Topics explored are recorded. The CTA engine decides whether this response should include a tour invite or application suggestion. All side effects — notifying staff of an escalation, sending a WhatsApp document — happen automatically without blocking the response.
After 3 sessions, intent score 38, topics explored: fees + boarding + Arabic support → CTA engine recommends embedding a tour invite in the next response.

When the AI steps back and hands to a human

The AI is honest about its limits. There are topics it will never attempt — it immediately connects the parent to the admissions team instead.

Always escalated to human Questions about specific students or their application status · Fee exceptions or negotiations · Complaints about specific teachers or staff · Custody or legal family situations · Medical advice or SEN diagnosis · Visa situations requiring legal guidance · Any topic the school marks as high-risk in the knowledge base
What the AI does instead of escalating wrong "That's a great question — I want to make sure you get the most accurate answer on that. Our admissions team can help directly. You can reach [Name] at [WhatsApp] right now." The school contact details are always included. The parent is never left stranded.
Who uses this system

Five types of users — each sees exactly what they need

The system is one platform with different views for different people. Parents see only the chat. Staff see the pipeline and timelines. No one sees more than they should.

👩‍👧
Parent
Prospective family
Sees only the chat interface — on website, Social media, or tablet. Never sees a dashboard, pipeline, or backend. The experience is purely conversational.
👩‍💼
Admissions staff
2–3 people
Pipeline board, parent timelines, AI action queue, document review, decision engine. Full write access. The AI tells them who to contact next and why.
👨‍💼
Principal
Read + approve
Analytics dashboard — conversion funnel, seat occupancy by programme, AI insights, campaign performance. Approves final admission decisions.
👩‍🏫
Teacher
Post-enrollment only
Student snapshot for incoming students — strengths, learning background, special notes. Read-only. Available only after enrollment is confirmed.
🛠️
AdminAdvisor team
Platform management
KB review dashboard, AI quality monitor, fact lock audit log, school onboarding tools. Super-admin access across all schools in the SaaS platform.
The build plan

6 months to launch — phased delivery

The system is built in three phases so IMAS can start seeing results before everything is complete. Phase 1 delivers the most critical piece first — the AI chat advisor. Everything else builds on top of it.

1
Months 1–2
Foundation — KB intake form + backend scaffold
IMAS fills in the 21-section knowledge base using the intake form. AdminAdvisor team reviews and approves each section.
Database schema set up — parent sessions, conversation history, KB sections, audit logs. Multi-tenant structure so other schools can be onboarded later.
KB sections vectorised — narrative content converted to embeddings for semantic search.
AI model abstraction layer built — swappable between Claude and GPT-4o without rewriting the engine.
Section-key mapping table built — resolves the naming gap between KB form keys and engine topic names.
Deliverable: Knowledge base 90% complete and approved. Backend infrastructure live on Supabase.
2
Months 2–4
Core engine — AI advisor live on website + Social media
Conversation engine built: semantic classifier, context builder, persona detector, CTA engine, retrieval layer, response generator (3 paths), fact lock layer, output router.
Session and identity layer — Tier 1 (anonymous by device), Tier 2 (WhatsApp number), Tier 3 (named). Merge logic for when parent gives contact.
Language detection — EN/BM/Arabic, including mixed-language messages. Session language memory.
Website chat widget embedded on IMAS website. Streaming SSE responses for real-time feel.
WhatsApp Business / Social Media API integration via BSP provider. Inbound and outbound routing. Document sending. PDPA consent flow on first message.
Walk-in tablet kiosk — PWA version of the chat widget, large-format UI for reception desk.
Deliverable: AI advisor live on website and Social media. IMAS parents can use it. All responses fact-locked.
3
Months 4–6
Staff tools — pipeline, timelines, analytics, and quality monitoring
Staff dashboard — pipeline board (Kanban by trust tier), individual parent timelines, AI-recommended action queue, document collection view, decision engine with human approval.
Analytics dashboard — conversion funnel, seat occupancy by programme (day vs boarding), drop-off analysis, campaign performance, AI insights.
KB review dashboard — AdminAdvisor team uses this to review school submissions, add AI answer guides, set risk levels, approve sections, trigger re-vectorisation.
AI quality monitor — weekly review of fact lock violations, blocked responses, and audit log. Identifies KB gaps and improves accuracy continuously.
Progressive application form — chat-history pre-fill, multi-step, auto-saves, statutory sign at school visit only.
Post-enrollment onboarding — class assignment, teacher snapshot, parent welcome sequence, fee reminders.
Deliverable: Full admissions module live. Staff replace spreadsheets. Ready to onboard second school.
Risk management

Known risks and how they are handled

Risk Likelihood Impact How it is handled
AI gives wrong fee figure
Parent makes financial decision on incorrect data		 Low High Fact lock validates every number before delivery. All fees are stored as exact records. Approximation language blocked at engine level. Audit log captures every response.
KB not 90% complete at launch
AI gives incomplete answers to live parents		 Medium Medium 90% gate enforced — AI runs in preview mode with a parent-visible notice. AdminAdvisor team supports school through KB completion in Month 1.
WhatsApp BSP approval delays
WhatsApp Business API can take weeks to approve		 High Medium BSP onboarding started in Month 1 (parallel to KB work). Website chat widget and tablet kiosk go live without WhatsApp. WhatsApp added in Month 3 when ready.
Parent shares sensitive information
Custody issues, medical conditions, legal matters		 Medium High Escalation detector triggers immediately. AI stops conversation and connects parent to human admissions team. Sensitive data flagged in session but not processed by AI.
PDPA / GDPR non-compliance
GCC parents covered by GDPR as EU data protection applies internationally		 Low High Consent flow triggered on first message. Data retention limits enforced by platform. Right to erasure handled. Supabase data residency in Singapore region.
AI sounds generic — not like IMAS
Defeats the purpose of a school-specific advisor		 Medium Medium Mitigated by KB Sections 1 and 19 — school identity, brand promise, advisor persona, and tone governance. AdminAdvisor team completes Tabs B-E (AI answer guides) per subsection before launch.
The bigger picture

IMAS is the first school. This platform serves many.

Everything built for IMAS is designed to scale. Every school gets its own isolated data space — they can never see each other's parent data, conversations, or knowledge bases. The platform layer handles compliance, AI model management, and billing centrally so each new school onboards faster than the last.

Multi-tenancy
Each school is isolated by school_id
Row-level security means School A can never access School B's data. The platform uses Supabase's built-in RLS policies to enforce this at the database level — it is structural, not just a permission checkbox.
AI model flexibility
Claude today. Any model tomorrow.
The AI model abstraction layer means swapping from Claude to GPT-4o, or using different models for different schools, requires changing a config value — not rewriting the engine. Embeddings always use OpenAI text-embedding-3-small for pgvector compatibility.
KB onboarding time
IMAS: 4 weeks. School 2: 2 weeks.
Each school shares the same KB intake form structure. The AdminAdvisor team's review and tab B-E completion process becomes faster with every school. The engine, the fact lock, and the retrieval layer are already built — only the knowledge changes per school.
What success looks like

Metrics we will track from day one

<3s
Average first response time
Was: 2–3 working days
70%
Tier 2 contact capture rate
Est. from trust ladder model
80%
Application form pre-fill rate
From conversation history
0
Wrong fee figures delivered
Fact lock target: zero violations
Part 10 — Interactive UI mockups

Screen-by-screen design

Five key screens that make up the day-to-day experience — for parents, for admissions staff, and for school leadership. Click each tab to explore.

imas.edu.my/admissions
A
Admissions Advisor
Responds instantly · Any language
EN
Hi! I'm the admissions advisor for IMAS. How can I help you today?
Just now
How much is boarding for Form 1?
Boarding fees for Form 1 are RM 28,500 per year (or RM 9,500 per term). This includes accommodation, meals, laundry, and all pastoral care. Day programme for the same year is RM 14,200 annually.

Would you like me to send you the full fee schedule with a breakdown of what's included?
3 seconds
Yes please, can you send the fee schedule?
Of course — shall I send it to your WhatsApp so you have it handy? What number should I use?
2 seconds
↑ Tier 2 — contact earned
English Bahasa Malaysia 中文
How the AI advisor works
Auto language detection
The AI matches the parent's language from their first message — English, BM, or Mandarin — with no language selection screen.
Programme-aware
The AI knows all fee tiers, boarding vs day differences, age group entry requirements, and scholarship criteria for every programme.
Session memory
A returning parent picks up exactly where they left off — 3 months later. No "how can I help you?" again.
Trust ladder built-in
The AI never demands contact. When the parent asks for something deliverable, it naturally asks for their WhatsApp — one field, fair trade.
admin.imas.edu.my / leads / amirah
AN
Puan Amirah Norzaidi
First contact: 14 Jan 2026 · 87 days in pipeline · Score: 78
Hot lead Grade 5 · Day programme
AI recommended action
Puan Amirah attended the virtual tour on 28 March but has not responded to the application link sent 5 days ago. She previously asked about scholarship eligibility. Suggest: send scholarship criteria PDF and invite to in-person campus visit on 19 April.
2 Apr
Application link sent by AI
WhatsApp message sent automatically after virtual tour. Link opened but not completed.
28 Mar
Virtual tour attended
Attended 45-minute virtual tour. Asked 3 questions: scholarship criteria, transport routes, co-curricular options.
12 Mar
Contact captured — Tier 2
Requested fee schedule PDF. Gave WhatsApp number +60 12-XXX XXXX. Anonymous profile merged to named lead.
14 Jan
First anonymous contact
Browsed from mobile. Asked about boarding fees and scholarship eligibility in English. AI answered fully. No contact collected.
Why staff love the timeline view
Full conversation history
Every message, document, and touchpoint from day one — including anonymous sessions before the parent gave their contact details.
AI next-action recommendation
The AI reads the full history and tells staff exactly what to do next and why. Staff approve or adjust — they never start from zero.
One-click approve & send
The AI drafts the message. Staff review it in seconds. One click sends it via WhatsApp. The timeline updates automatically.
2–3 staff, 150+ leads
Because the AI handles triage and next-action logic, a small admissions team can manage a large pipeline without spreadsheets or missed follow-ups.
apply.imas.edu.my / grade-5 / step-2
Admission application — Grade 5, Day programme
Step 2 of 4 — Child details
● Saved
Pre-filled from your conversation
Already known — from your conversation with our advisor
Grade applying for
Grade 5
From conversation
Programme
Day
From conversation
Current curriculum
National (SK)
From conversation
Your WhatsApp
+60 12-345 6789
Verified contact
Please complete — 4 fields
Child's full name (as per IC)
Date of birth
IC / birth certificate number
Any known medical conditions?
Documents — upload or send via WhatsApp
✓
Birth certificate
Received via WA
○
Latest school report card
Pending — WhatsApp +60 12-345 6789 or upload below
Why this form looks different
60–80% pre-filled
The AI has been in conversation with this parent for 3 months. It already knows the grade, programme, curriculum, and contact details.
Documents via WhatsApp
Parents can send birth certificates and report cards as WhatsApp photos. OCR extracts and validates the data automatically.
Saved automatically
Progress saves every field. A parent can close and return weeks later — the form remembers exactly where they left off.
Statutory declaration
The one legally required signature is handled as a single printout signed at the campus visit. Not a blocker to completing everything else digitally.
admin.imas.edu.my / pipeline
Admissions pipeline — April 2026
153 active leads · 41 enrolled · Day: 24 seats left · Boarding: 6 seats left
Anonymous38
Device #8821
Asked: boarding fees, Form 1
Intent: boarding · 3 sessions
Device #9103
Asked: scholarship Grade 7
Intent: day · 1 session
+ 36 more devices
Soft contact47
+60 12-XXX 4421
Fee schedule sent 3 Mar
Score: 62 · Day
Puan Ros (+60 19-XXX)
Brochure sent 28 Feb
Score: 71 · Boarding
+ 45 more leads
Named leads31
Puan Amirah
Hot · 78
Grade 5 · Day · 87 days
AI: send scholarship info
Pn. Faridah Hassan
Grade 3 · Boarding · 42 days
Score: 65
+ 29 more leads
In application18
Mr. Tan Wei Liang
Grade 7 · Boarding
Form 60% done
Dr. Priya Nair
Grade 1 · Day
Form 85% done
+ 16 more applicants
Enrolled41
Marcus Lew
Grade 3 · Day · 6-month journey
Enrolled ✓
Nur Izzati Ramli
Grade 7 · Boarding · 4-month journey
Enrolled ✓
+ 39 enrolled
Pipeline shows anonymous device IDs until contact is captured — no data is lost at any tier
How the pipeline board works
Zero data loss from day one
Anonymous devices are tracked from their first visit. When they give their contact later, everything is merged. No lead disappears.
Intent scores, not guesswork
Each lead has an AI-calculated intent score (0–100) based on topic depth, session count, and engagement signals. Staff know who to prioritise.
Programme-level seat tracking
Filter by programme to see exactly how many seats are left and which grade levels are filling fastest. Critical seats surface automatically.
AI next action on each card
For every named lead, the AI suggests the next action. Staff see it at a glance — no need to open the full timeline unless they want to.
admin.imas.edu.my / analytics
Intelligence dashboard — April 2026
Live · Auto-refresh every 5 min
153
Active leads
+22% vs March
41%
Conversion rate
+8pp vs target
6
Boarding seats left
Critical — Grade 7
3.2s
Avg. first response
Was: 26 hours
Conversion funnel — all programmes
Anonymous
153
Soft contact
112
Named leads
79
Applied
58
Enrolled
41
Seat forecast — by programme
Day — Grade 5
4 seats
Full in 3 weeks
Boarding — Grade 7
2 seats
Open waitlist now
Day — Grade 3
12 seats
On track
AI insights
!
Boarding Grade 7 critical: 2 seats remain with 6 active applications. Recommend opening waitlist today and expediting 3 pending decisions.
↩
Re-engagement opportunity: 14 soft-contact leads last active 30+ days ago. Personalised nudge expected to recover 3–5 leads.
↑
Language insight: BM-speaking leads convert at 1.8× the rate of English-speaking leads for the day programme.
→
Form completion: Application completion rate improved from 38% to 81%. Average time: 8 minutes (was 45 minutes).
What leadership sees every morning
Live funnel, not a monthly report
The conversion funnel updates in real time. Leadership can see exactly how many leads are at each stage and where drop-off is happening.
Seat forecasting by programme
The system projects when each grade and programme will fill based on current pipeline velocity. Critical seats surface automatically with recommended actions.
AI-generated insights
The AI reads the full pipeline and surfaces specific, actionable observations — not just numbers. "Open the boarding Grade 7 waitlist today" not "boarding is popular."
Language & channel breakdown
Understand which languages are converting, which channels are driving the most qualified leads, and where to focus outreach for the next intake period.