Case Study

Agentic AI Tutor Speeds up GAMSAT Learning Curve and Automates Essays Review

10 minutes to review unlimited number of essays

Full automation of student’s essay evaluation

AI Agent as a “digital twin” of a tutor

High-quality feedback with insights on essay improvement

Human-like assessment with 100% accuracy and quality

Services provided

Agentic AI custom software development

Frontend Development

UX/UI design

Backend Development

About the Project

Client
Educational Software
Company
Industry
E-learning,

Tutoring
Location
Australia
Company Size
14 employees
Duration
October 2024 – present

An Australian-based EdTech and mentoring service that helps applicants prepare for an admission exam—GAMSAT, mandatory to enter medical universities in Australia, the UK, and Ireland. IT provides courses, practice exams, study materials, and AI-powered software for self-learning.

Challenges

When the online mentoring platform picked up steam among students who sought ways to master three-section GAMSAT, the client found it difficult to match the growing demand. Plus, mentors had to deal with constraints of the standard systems used for the preliminary review of essays.

Time-consuming manual overhead

Highly skilled PE agents spent excessive time on routine tasks like document processing, data entry, instead of strategic analysis.

Approximate evaluation

Highly skilled PE agents spent excessive time on routine tasks like document processing, data entry, instead of strategic analysis.

Fragmented feedback

Highly skilled PE agents spent excessive time on routine tasks like document processing, data entry, instead of strategic analysis.

The goal of our client, as an expert in GAMSAT mentoring, was to create and integrate an AI tool into the current online web platform that would make preparation for the exam affordable for all applicants and help increase the web service user base and its conversion rate.

Solution

Agent Architecture

OpenAI Core

Relying on OpenAI models, this AI agent has deep contextual understanding and advanced reasoning. Fine-tuned on real essay samples and expert feedback, it can analyze complex writing attributes and identify the slightest weaknesses.

Orchestration Engine

Langgraph and LangChain frameworks allow the AI Agent to switch seamlessly between different tools, aligned with the essay’s requirements. This engine maintains the context across multiple evaluation steps and ensures the output replicates the tutor’s distinctive style.

Agent’s Personality

To create a human-like feedback and emulate the real tutor’s personality, we calibrated the system upon the mentor’s approach and language. This type of personalization makes the AI tutor’s commentary precise, engaging, and trustworthy.

Toolset Integration

The AI agent can effortlessly access a supporting suite of integrated tools to enhance its evaluation process. Among them are:

Custom RAG System

A Retrieval-Augmented Generation (RAG) pipeline enriches the LLM’s input with additional context from the curated tutor’s books, courses, and educational videos for more informed feedback.

Essay Metrics Extraction Tool

This module automatically parses text to extract key metrics such as structure, coherence, clarity, argument quality, idea development, and overall composition. Using those, it measures essay performance to provide holistic feedback.

Grammar and Style Analyzer

Leaning on NLP algorithms, this tool reviews language usage, grammar, and stylistic elements, pinpointing areas for improvement.

Plagiarism Checker

The integrated tool scans the essay against a database of sources, highlighting issues and ensuring it adheres to the academic standards.

Feedback Synthesis Module

It generates comprehensive, actionable, and expert feedback with strengths, areas for improvement, and recommendations, imitating the tutor’s review style.

Technologies

OpenAI
Langchain
LangGraph
RAG (Retrieval Augmented Generation)

Business Value & Outcomes

Developed as a dedicated GAMSAT tutor, the system automates the rigorous process of essay evaluation and speeds up the grading process. Among its benefits are:

Faster Learning Curve

By offering a real-time, in-depth analysis, the system lets students iteratively improve their essays—reducing turnaround times and enhancing learning outcomes.

Scalability

The AI Agent can review essays rapidly, delivering detailed, consistent evaluations. Thus, tutors can handle more essays with the same quality.

Cost Efficiency

It's a more accessible alternative compared to the real tutor’s evaluation. Students benefit from high-quality, personalized feedback at a lower price.

Results in up to 10 minutes

Unlike a tutor's review that often takes a week, students can get an assessment of as many essays as they can write in minutes. The more essays are written and processed before the exam, the better the final grade is.

Lawyers Focus on Results

Automation of routine evaluation tasks allows tutors to focus on high-level mentoring and innovative educational strategies.