Introducing AI Operating System for Business & School — our most powerful products yet.Explore products →
App Corp
Full-service software engineering
Engineering your experience…
App Corp
Full-service software engineering
Engineering your experience…
How we built a full-stack web + mobile EdTech platform with 6 purpose-built AI agents — replacing paper registers, WhatsApp groups, and Excel fee sheets with a cloud-native multi-tenant SaaS live across South Asia, the Middle East, and Africa.
About The Client
An AI-first school operating system that automates admissions, attendance, fees, and parent communication for K-12 schools across developing markets — purpose-built for schools in Pakistan, Nigeria, UAE, and Kenya.
EdTech SaaS
Internal Product
Global (South Asia, MENA, Africa)
Mid-Market
Technology Stack
The specific stack was chosen after evaluating the client's scalability requirements, compliance constraints, team capabilities, and long-term maintainability — not retrofitted to existing preferences.
Schools across South Asia, the Middle East, and Africa were operating on fragmented manual infrastructure — paper attendance registers, Excel fee sheets, WhatsApp parent groups, and disconnected legacy ERPs. The administrative overhead was crushing staff who should have been focused on education. Attendance registers caused daily data loss. Parent communication was scattered across WhatsApp groups with no audit trail. Fee tracking lived in spreadsheets with no automated reminders, so collections slipped. Admission inquiries were tracked in notebooks — leads fell through cracks with no pipeline visibility. Principals had no real-time view of attendance rates, fee collection status, or academic performance.
0
unified systems — everything ran on paper and WhatsApp
Key Pain Points
Paper attendance registers — daily data loss, no real-time visibility for administrators
WhatsApp parent groups — fragmented communication, zero audit trail, no structured updates
Excel fee sheets — missed payments, no automated reminders, manual reconciliation
Manual admission processing — leads tracked in notebooks, no follow-up automation
Zero operational visibility — principals blind to attendance rates, fee collection, performance metrics
Rather than building another school ERP with a thin AI veneer, we designed EduPilotPro from the ground up as an AI-first School Operating System. The key design decision: deploy purpose-built AI agents for each administrative function, not a generic chatbot bolted onto a legacy product. We built a Next.js admin dashboard for school staff, a React Native mobile app for parents, and a REST API powering both — all deployed on Railway with true multi-tenant isolation enforced at the database layer via Supabase Row-Level Security.
Five core modules — admissions, attendance, fees, academics, and parent communication — in a single unified interface. Each school operates in its own subdomain with full data isolation via Row-Level Security. Real-time attendance tracking, fee collection status, and admission pipeline visibility.
A cross-platform mobile app for parents covering real-time attendance notifications, fee payment and receipts, academic performance updates, and school announcements. Offline-first architecture with push notifications replacing WhatsApp as the primary school-to-parent channel.
Admission agent (auto-enrolment pipeline), attendance agent (natural language marking — 'All present except Ali who is absent'), fee reminder agent (automated WhatsApp + email reminders), academic structure agent (class/subject setup), role-based access agent (permissions per staff type), and reporting agent (automated performance summaries). Each runs on Google Gemini with WebLLM as a secondary fallback source.
Centralised school-to-parent notifications replacing WhatsApp group chaos — attendance alerts, fee due reminders, exam schedules, school events, and emergency broadcasts. Structured delivery with read receipts, multilingual support, and opt-in channels (WhatsApp, email, in-app).
Every engagement follows a disciplined, phased delivery methodology — built on transparency, measurable milestones, and continuous client collaboration throughout.
We designed the multi-tenant architecture first — every school in an isolated subdomain environment. The database schema was built around school operations: admissions pipeline states, attendance tracking with period-level granularity, fee structures with instalment plans, academic calendar entities, and parent communication preferences. Supabase Row-Level Security enforced tenant isolation at every query.
We built the five core modules in parallel — admissions, attendance, fees, academics, and communication. Each module was built as an independent feature set within the monorepo, sharing common primitives (tenant context, auth, notification dispatch) via internal packages. A staging environment with real school data from a pilot partner was live within four weeks.
The AI agents were built on Google Gemini with structured prompt chains for each administrative function. The attendance agent understood natural language input — 'All present except Ali who is absent' — and parsed it into structured attendance records. The React Native app was designed for the lowest-common-denominator devices common in target markets, with offline-first caching and optimistic UI for areas with unreliable connectivity.
Schools in Pakistan, Nigeria, UAE, and Kenya have distinct operational needs, fee structures, and communication patterns. We deployed with market-specific configurations — local payment gateways, curriculum-aligned academic structures, and language-localised parent communication. Each market deployment included remote training sessions for administrative staff and a 2-week parallel-run period alongside existing manual processes.
Every number below is a verified, real-world outcome. No hypotheticals, no projections — these are the results delivered.
6
AI Agents in Production
Admission, attendance, fee reminders, academic structure, role-based access, and reporting — each handling its workflow end-to-end without manual staff input
5
Core Modules Live
Admissions, attendance, fees, academics, and parent communication — covering the full school operation lifecycle in a single platform
3
Regions Deployed
Live across South Asia (Pakistan), Middle East (UAE), and Africa (Nigeria, Kenya) — each with market-specific configuration
True
Multi-Tenant Isolation
Each school operates in an isolated subdomain environment with Supabase Row-Level Security — zero data bleed between tenants
Web + Mobile
Full-Stack Delivery
Next.js admin dashboard for staff, React Native mobile app for parents, and REST API powering both — deployed on Railway
14 wks
Time to Launch
From first line of code to multi-region production deployment serving real schools with thousands of students
“We took fragmented, paper-driven school administration and turned it into a cloud-native, AI-first SaaS platform. The key insight was deploying purpose-built AI agents for each function — not a generic chatbot. Each agent handles its workflow end-to-end, from admission inquiry to fee reminder to attendance report. Schools that were drowning in paper and WhatsApp are now running on a single command centre.”
Taimoor Khan
Cloud Infrastructure Architect & AI Engineer · App Corp