Mentra AI – Adaptive Learning Intelligence Platform

Mentra AI -- Adaptive Learning Intelligence Platform

Inspiration

Learning today is highly fragmented. Students attend lectures, collectnotes, watch videos, solve problems, and interact with AI chatbots.However, these tools rarely understand what the student actuallyknows. Every conversation starts from scratch, recommendations aregeneric, and misconceptions often remain hidden until an exam.

What if AI could build a living understanding of every learnerinstead of just answering questions?

That question became Mentra AI.

Our vision was to create an AI mentor that continuously learns alongsidethe student, understands conceptual mastery, detects misconceptions, andpersonalizes learning---not just for one session, but throughout theentire learning journey.

The Problem

Current AI tutors are primarily reactive.

They answer questions only after students ask.

They forget previous learning progress.

They cannot measure conceptual understanding.

They provide identical explanations to different learners.

Teachers have little visibility into where students struggleconceptually.

This results in passive learning instead of true conceptual mastery.

Our Solution

Mentra AI transforms static study notes into an adaptive, personalizedlearning ecosystem.

Students upload their learning material, and Mentra AI automatically: -Extracts structured knowledge. - Generates interactive concept maps. -Creates a personalized concept graph. - Detects conceptual gaps. -Recommends adaptive learning paths. - Generates personalized quizzes. -Provides evidence-grounded AI tutoring. - Continuously updates thelearner's knowledge model after every interaction.

What Makes Mentra AI Different?

Mentra AI introduces a Personal Concept Graph, a continuouslyevolving representation of each learner's understanding. Everyinteraction strengthens mastered concepts, detects misconceptions, andupdates future recommendations, enabling personalized tutoring, adaptiveassessments, explainable recommendations, and continuous learninganalytics.

How We Built It

Document Understanding

Uploaded PDF notes are parsed into structured chunks.

Retrieval-Augmented Generation (RAG)

Answers are grounded using the uploaded learning material beforegeneration.

Concept Graph Generation

Relationships between concepts are transformed into an interconnectedknowledge graph powering prerequisite detection, misconceptionidentification, adaptive recommendations, and revision plans.

Adaptive Learning Engine

Every interaction updates the learner model by strengthening masteredconcepts, detecting weak areas, adjusting quizzes, and recommending nextlearning objectives.

Learning Analytics

Students receive mastery scores, progress visualization, conceptcoverage, and recovery plans. Educators receive actionable conceptualinsights.

Technology Stack

Frontend

React

TypeScript

Tailwind CSS

Vercel

Backend

FastAPI

Python

Render

Large Language Models

Retrieval-Augmented Generation (RAG)

Semantic Search

Concept Graph Generation

Adaptive Learning Engine

Challenges We Faced

Designing an adaptive learner model beyond a traditional chatbot,integrating a Vercel frontend with a Render-hosted FastAPI backend, andensuring AI responses remained grounded in uploaded notes rather thanrelying solely on model knowledge.

What We Learned

We learned how to combine retrieval systems, knowledge representation,adaptive learning, backend engineering, deployment pipelines, and userexperience into a unified educational AI platform.

Impact

Mentra AI can support university students, competitive exam aspirants,self-paced learners, corporate training, and professional certificationprograms by enabling personalized, concept-driven learning.

Future Roadmap

Multimodal learning

Long-term learner memory

Predictive academic risk detection

Institution-wide learning analytics

AI teaching assistants

Our goal is to transform AI from a system that answers questionsinto one that truly understands every learner.