How AI Is Helping 10,000+ Indian Students Prepare for NEET and JEE for Free

Source: Dev.to

The Problem We Built For

In 2010, I was tutoring a brilliant 17‑year‑old from Indore. She had the conceptual grasp to crack NEET, but lived 800 km from the nearest quality coaching center. Her parents couldn’t afford the 3‑4 lakh rupees for online coaching subscriptions. She practiced from old PDFs, had no way to validate her answers in real time, and when doubts hit at 11 PM, she was stuck.

That student didn’t make it that year. She could have.

After 14 years of mentoring 500+ students across tier‑2 and tier‑3 Indian cities, the pattern was undeniable: talent isn’t distributed by geography or income. Access to quality preparation is.

So in 2023, I decided to build EnrollAI—a free, AI‑powered Telegram bot for NEET and JEE aspirants. Today, 10,000+ students use it daily. This is how it works and why it matters.

Why Telegram? Why Free?

We chose Telegram because:

• 80 % of NEET/JEE aspirants already use it for study groups
• No app installation barrier — just click a link
• Zero data consumption for text‑based queries in many regions
• Built‑in for accessibility — works on 2G networks

Free wasn’t a PR move. It was intentional. The students we wanted to help often can’t afford subscriptions. If cost was a barrier, we’d be solving the wrong problem.

How EnrollAI Actually Works

1. Adaptive MCQ Practice

Students get randomized questions from our database of 10,000+ curated MCQs across:

• Physics, Chemistry, Biology (NEET)
• Physics, Chemistry, Math (JEE)

The algorithm works like this:

• First attempt: Questions span all difficulty levels
• Second attempt: The AI weights questions based on your performance
• Weak topics get higher frequency (spaced repetition)
• Strength in a topic (e.g., organic chemistry) → fewer easy questions there

In three months, an average user completes 300‑500 questions—equivalent to 12‑20 full‑length NEET papers.

2. Instant Doubt Solving

When a student submits a doubt:

1. The query is parsed for subject and topic.
2. Our AI retrieves relevant conceptual notes from our database.
3. It generates a step‑by‑step solution with visual pointers.
4. It explains the most common mistakes for that question type.

Example: A student from Nashik asked about hybridization in VSEPR theory at 9:47 PM. Within 90 seconds, they received a detailed explanation—previously they would have waited up to 48 hours for an online tutor.

3. Progress Tracking

Students see:

• Accuracy percentage across each subject
• Topic‑wise performance heatmaps
• Consistency metrics (practice regularity)
• Estimated readiness score based on mock performance

When a student from a tier‑3 city sees improvement from 58 % to 73 % in inorganic chemistry, it’s more than a number—it’s evidence that consistent effort works.

The Real Impact

After 18 months, we’re seeing:

• 7,400+ students have completed at least 50 practice questions
• Average daily active users: 2,100 across all time zones
• Doubt resolution rate: 87 % (students report the solution helped them understand)
• Geographic spread: 28 states and UTs; 34 % from tier‑2/3 cities

A survey of 800 active users revealed that 62 % couldn’t afford coaching centers, relying on YouTube, old textbooks, and peer groups. EnrollAI became their accessible tutor.

“She studied with your bot from her home, asked doubts at any time, and it never made her feel alone.” — Mother of a student from Aurangabad who scored 618 on NEET (above the cutoff for government medical colleges).

The Technical Reality

EnrollAI isn’t magic. It consists of:

• GPT‑4 integration for doubt solving (API costs covered; students never pay)
• PostgreSQL backend storing performance data with strict privacy protocols
• Custom NLP pipeline to extract subject and topic from unstructured doubts
• Retrieval‑Augmented Generation (RAG) to source solutions from validated textbooks and past papers, avoiding hallucinations

We spent 200 hours ensuring accuracy. Every explanation is fact‑checked against NCERT standards and expert validation. An AI‑generated wrong answer in a medical entrance exam isn’t a learning opportunity—it’s dangerous.

Why This Matters for Developers

If you’re building EdTech or AI products, here’s what we learned:

• Free isn’t unsustainable if you have a mission. We’re funded by donations and earn revenue from advanced features (coming next quarter) for students who can afford them.
• Local context > global optimization. Designing for “students in India” is useless. Design for “students with 2G networks, studying at midnight because they work during the day, speaking 8 regional languages.”
• Privacy is non‑negotiable in education. We don’t sell data. Ever. It’s the only competitive advantage that matters when you’re building for trust.
• Offline‑first architecture saves lives. We cache previous doubts and solutions locally. When the internet cuts (common in rural areas), students can still study.