Skip to main content
Back to Home Flagship Product Case Study

ThinkGrades

Building an AI-native operating system for modern schools.

A production SaaS case study demonstrating how Tectrom handles complex workflows, multi-tenant security, role-specific UX, AI integration, and continuous product development.

Multi-tenant SaaS Parent & Teacher PWAs AI Diagnostic Copilot PDF Marksheet Engine Fee Ledgers PostgreSQL RLS
Visit thinkgrades.in
Project Overview & Engineering Facts
Flagship Product thinkgrades.in ↗
Ownership & Role Built & Operated by Tectrom
Core Architecture Multi-Tenant PostgreSQL RLS
Current Status ● Live Production Operating
Live Production SaaS
ThinkGrades main institutional administrative dashboard overview showing live school analytics.
AI Morning Briefing

98.4% daily attendance logged across 12 grades. 3 fee follow-ups automatically dispatched.

Performance Trend Term 2 Analytics
Overall Average: 84.2% (+3.1%)

Admin Dashboard • Central institutional hub showing real-time metrics, system health, and AI briefings.

Origin Story

Why we built ThinkGrades.

How an AI chatbot for schools evolved through real customer feedback into a complete production School Operating System.

ThinkGrades didn’t begin as a School ERP. It started with a simple question between two friends: what if school leadership, teachers, parents, and students could ask an intelligent assistant anything about their institution and get an instant, reliable answer?

The original vision was focused entirely on conversation. We set out to build an AI chatbot that could answer operational queries on demand—checking daily attendance percentages, querying exam performance across classes, retrieving outstanding fee statuses, looking up homework assignments, and summarizing student progress feeds.

While building the initial prototype, however, we ran into a fundamental engineering constraint: an AI assistant is only as good as the underlying data behind it.

We quickly discovered that school data was deeply fragmented. Critical records didn’t live in structured databases—they were scattered across ad-hoc Excel spreadsheets, physical paper registers, WhatsApp groups, legacy desktop software, and unwritten staff routines. There was no single, trustworthy source of truth.

To make the chatbot genuinely useful, we had to build the data platform underneath it first. We engineered core modules for student rosters, daily attendance, exam scheduling, multi-part marks entry, tuition fee ledgers, parent communication feeds, and class analytics. Initially, every single one of these modules was built for a single purpose: acting as a clean, reliable data pipeline for the AI.

Then something remarkable happened. Once schools started using the platform, they began requesting complete ERP capabilities. Teachers and administrators didn't just want to query the AI assistant—they wanted to manage their entire daily operations inside the same unified system.

Gradually, ThinkGrades evolved from a lightweight AI chatbot into a full-fledged School Operating System. Today, AI is no longer the entire product. It has become the intelligence layer running across the platform—helping administrators automate workflows, detect academic score variances, generate daily morning briefings, and make better decisions.

“We learned that intelligence cannot substitute for structure. Before an AI assistant can answer complex operational questions, you must first build the operational source of truth underneath it.”

— The ThinkGrades Product Engineering Team

Three Principles That Shaped the Product

Principle 01

AI Starts with Good Data

Intelligence is a direct function of data quality. Without a clean relational database foundation, AI models generate hallucinations rather than trustworthy operational answers.

Principle 02

Build Foundation Before Intelligence

Software must solve core daily utility first. High-frequency workflows like attendance, marksheets, and fee ledgers earn user trust long before AI insights add leverage.

Principle 03

Customer Feedback Shapes Products

Great products are rarely manufactured in isolation. The shift from a simple chatbot to a production School OS happened naturally by listening to how schools actually work.

Product Evolution Timeline
💡 Idea
AI Chatbot for Schools

Conversational assistant answering school queries.

📊 Realization
AI Needs Structured Data

Discovered school data was scattered & unorganized.

🏗 Foundation
Built Core Platform

Engineered rosters, attendance, marks & fee ledgers.

🏫 Feedback
Requested Full ERP

Schools requested complete daily operations inside OS.

🚀 Today
School Operating System

Production-grade AI-native School Operating System.

This journey shaped how we approach product strategy for institutional clients: rather than building isolated tools, we build unified operating systems. Here is how we turned that origin story into concrete product strategy.

Platform Scope

Product at a Glance overview.

A quick overview of the platform we built and continue to evolve.

AI-Native Platform

AI is integrated throughout the platform to assist with insights, reporting, and intelligent workflows rather than existing as a standalone feature.

Multi-Tenant SaaS

A single platform designed to securely support multiple schools through isolated data and role-based access.

Role-Based Experience

Dedicated experiences for administrators, teachers, parents, accountants, and platform administrators.

Parent & Teacher PWAs

Mobile-first Progressive Web Apps built for high-frequency daily workflows.

Academic Operations

Attendance, examinations, marks, report cards, grading, homework, and classroom management in one connected platform.

Financial Management

Fee structures, invoices, payments, receipts, collection tracking, and reporting.

Analytics & Insights

School, class, and student analytics supported by contextual AI-powered insights.

Modern Architecture

Built using Next.js, React, Supabase, PostgreSQL, Tailwind CSS, and deployed on modern cloud infrastructure.

Unified System

Designed for modern schools.

ThinkGrades combines administration, academics, finance, communication, and AI into one connected operating system, reducing the need for disconnected tools and manual workflows.

Institutional Architecture Flow
Students Teachers Parents
ThinkGrades OS Engine
Attendance Fees Exams Analytics AI Intelligence
Product Strategy

From fragmented legacy tools to an active School OS.

Educational institutions often run core daily workflows across spreadsheets, paper registers, legacy ERP software, and disconnected messaging apps.

Traditional School ERP

Legacy Model
  • Stores static records in passive database tables
  • Form-heavy administrative entry screens
  • Siloed, disconnected modules requiring re-entry
  • Delayed reporting generated weeks after terms end

ThinkGrades School OS

Active OS Model
  • Connected, event-driven academic workflows
  • Role-specific experiences (Admin, Teacher PWA, Parent PWA)
  • Automated mark processing & fee receipts
  • Real-time insights and AI-assisted decisions
What We Built

Production-grade SaaS modules & real product screenshots.

A deep look into the core workflow modules designed, engineered, and operated within ThinkGrades OS.

Module 01 Core Setup • School Operations

School Operations & Roster Management

Centralized administration for academic sessions, sections, subjects, student records, teacher allocations, and multi-tenant role permissions.

✓ Student List & Profiles ✓ Class & Section Mapping ✓ Subject Allocation ✓ Role Security Controls
Illustrative Product UI — Anonymized Sample Data
ThinkGrades student management dashboard displaying student lists, profile records, and class allocations.

Student Management • Centralized student list, profiles, and institutional academic roster setup.

Illustrative Product UI — Anonymized Sample Data
ThinkGrades attendance management dashboard showing daily classroom attendance, attendance heatmaps, and institutional reports.

Attendance Dashboard • Daily attendance tracking with heatmaps, reports, and instant alerts.

Module 02 Mobility & Tracking • Attendance

Daily Attendance & Mobile PWAs

Mobile-first Progressive Web Apps engineered for one-tap classroom marking by teachers and real-time attendance visibility and leave applications for parents.

✓ Daily Classroom Attendance ✓ Attendance Heatmaps ✓ Automated Parent Alerts ✓ Leave Range Management
Module 03 Assessment Engine • Exams & Marksheets

Exams, Marks Entry & Server PDF Marksheets

Multi-component evaluation supporting theory, oral, and practical weightages with automated calculation engines and headless PDF marksheet generation.

Illustrative Product UI — Anonymized Sample Data
ThinkGrades exam management screen displaying exam schedules, grading scales, and assessment configurations.

Exam Management • Exam setup & grading scale configuration.

Illustrative Product UI — Anonymized Sample Data
ThinkGrades marks entry interface showing theory, oral, and practical subject mark recording.

Marks Entry • Multi-component subject score entry grid.

Module 04 Financial Operations • Fee Management

Fee Management & Collection Ledgers

Streamlined tuition fee structures, collection recording, automated invoice generation, outstanding payment tracking, and financial analytics.

✓ Instant Collection Recording ✓ Digital Invoice Receipts ✓ Outstanding Fee Ledgers ✓ Collection Revenue Analytics
Illustrative Product UI — Anonymized Sample Data
ThinkGrades fee management dashboard displaying collection metrics, invoice records, and outstanding fee ledgers.

Fee Management • Fee collection, digital invoices, and outstanding payment analytics.

Illustrative Product UI — Anonymized Sample Data
ThinkGrades institutional analytics and AI diagnostic dashboard displaying performance charts, cohort insights, and morning briefings.

Analytics & AI • Performance charts, diagnostic insights, and automated daily AI briefings.

Module 05 Intelligence • Analytics & AI

Institutional Analytics & AI Briefings

Real-time performance distribution charts, cohort score trend tracking, and automated AI diagnostic briefings providing actionable recommendations to administrators.

✓ Class Performance Distributions ✓ Historical Score Variance ✓ AI Morning Executive Briefings ✓ Cohort Diagnostic Insights
Technical Rigor

3 Hardest engineering challenges solved.

A detailed look into the architectural trade-offs, security models, and database challenges we solved while engineering ThinkGrades.

Challenge 01

Secure Multi-Tenant Row Level Security (RLS)

The Engineering Hurdle:

Multiple institutions share the SaaS application database infrastructure, requiring strict tenant isolation to prevent cross-school data leakage.

Our Architectural Solution:

Implemented school-based tenancy utilizing PostgreSQL Row-Level Security (RLS) policies in Supabase. Database queries evaluate active school membership context, enforcing tenant isolation at the database layer.

  • Row-Level Security (RLS) policies enforced at database layer
  • Explicit user-school membership mapping
  • Super-admin tenant context switching
PostgreSQL Supabase RLS TypeScript Next.js
Challenge 02

Complex Academic Data Modeling

The Engineering Hurdle:

School marksheets vary greatly across institutions and subjects—mixing theory, oral, practical components, and custom grading boundaries.

Our Architectural Solution:

Engineered a flexible relational schema modeling marks as multi-part assessment components with dynamic weightages rather than hardcoding flat scores.

  • Dynamic theory, oral, and practical component structures
  • Configurable grading scales (letter grades, GPA, percentages)
  • Flexible exam classification and term aggregations
PostgreSQL Schema Relational Modeling TypeScript Engine
Challenge 03

Server-Side Report Card PDF Generation

The Engineering Hurdle:

Generating end-of-term student marksheets requires consistent printable layouts and reliable batch generation across entire classes.

Our Architectural Solution:

Engineered a dedicated server-side Puppeteer PDF service rendering HTML/CSS templates into downloadable marksheets.

  • HTML/CSS template-to-PDF rendering
  • Batch marksheet generation queue
  • Scalable worker service deployment
Puppeteer PDF Service Node.js Railway Infrastructure
Infrastructure Spec

System architecture & technology stack.

Clean layered architecture connecting user interfaces, application routes, security boundaries, database tables, and external microservices.

System Architecture Flow

End-to-End Data Pipeline

Role Interface Admin Dashboard
Mobile PWA Teacher App
Mobile PWA Parent Portal
Next.js Application Layer Sub-Second Route Handlers & Edge Rendering
Security Boundary Authentication + Row Level Security (RLS)
Database Layer Supabase PostgreSQL (Multi-Tenant Schema)
AI Diagnostics OpenAI API
PDF Engine Marksheet Service
Notifications Web Push VAPID

Frontend & Apps

  • Next.js App Router
  • React
  • Tailwind CSS v4
  • Teacher & Parent PWAs

Data & Security Layer

  • Supabase PostgreSQL
  • Row Level Security (RLS)
  • Strict Schema Migrations

AI Layer

  • OpenAI API
  • Vercel AI SDK
  • Structured JSON Output

Infrastructure & APIs

  • Vercel Edge
  • PDF Generation Service
  • Web Push / VAPID
UX & Intelligence

Product decisions & practical AI integration.

Key trade-offs made during product development to reduce friction, alongside practical AI assistance.

Decision 01

Why a dedicated Teacher PWA?

Teachers repeatedly perform a small number of high-frequency actions (attendance, homework, marks entry). Designing a dedicated PWA eliminated desktop clutter and enabled fast mobile completion.

Decision 02

Why simplify the Parent experience?

Parents need clear visibility into their child’s progress and fee receipts—not administrative software. We kept the parent PWA focused entirely on progress feeds and marksheets.

Decision 03

Why role-based interfaces?

Showing only functionality relevant to an active user role reduces cognitive load, speeds up task completion, and limits accidental administrative changes.

Decision 04

Why multi-tenancy from Day 1?

ThinkGrades was designed as a scalable SaaS product rather than separate software installations per school, dramatically simplifying maintenance.

Contextual AI Assistant Practical AI Use Cases (No Novelty Fluff)
● [LIVE PRODUCTION] Student Variance Diagnostics

Identifies score variance for early teacher intervention.

● [LIVE PRODUCTION] Class Analytical Summaries

Aggregates subject performance overviews for admins.

● [PRIVATE BETA] CSV Import Column Mapper

Assists admins with legacy student data ingestion.

Product Evolution

Current status & engineering roadmap.

Clear differentiation between deployed production modules, beta features, and planned engineering initiatives.

● LIVE PRODUCTION

Core School OS

Multi-tenant RLS, operations, attendance, marks, fees & PWAs.

● PRIVATE BETA

API Rate Limiting

Enhanced rate limiting and tenant security auditing.

● IN DEVELOPMENT

Async PDF Queue

Worker queue for large-scale batch marksheet generation.

● PLANNED ROADMAP

Offline PWA Sync

Background sync for attendance in low-connectivity environments.

Engineering Wisdom

What building ThinkGrades taught us.

Three core lessons learned while engineering multi-tenant SaaS platforms.

Lesson 01

Complex workflows need simple interfaces.

Complexity should live inside the database and backend architecture—never on the user's screen.

Lesson 02

Different users require different experiences.

An administrator desktop dashboard cannot simply be shrunk down and called a mobile app for teachers or parents.

Lesson 03

Architecture decisions become product decisions.

Multi-tenancy, Row Level Security, and clean schema design directly determine how fast future features can be built.

Product Engineering Partnership

Building a multi-tenant SaaS platform or replacing a fragmented workflow?

Talk directly with the product engineering team behind ThinkGrades. 30-minute technical discussion • No sales handoff.

Visit thinkgrades.in