Entexams.ai – AI-Driven Study Companion
Entexams.ai is an education-focused startup aiming to build a fully AI-driven study companion for students preparing for competitive entrance exams. Their vision was to bring together the best capabilities of modern AI models and deliver them as a single, unified learning platform.
The Challenge
Students preparing for entrance exams often struggle with scattered resources, inconsistent study plans, and lack of personalized guidance.
The client wanted a platform that could:
- Combine the strengths of multiple AI platforms (ChatGPT, Gemini, Claude, Perplexity)
- Offer module-based mentorship instead of generic chat responses
- Stay cost-effective despite using multiple LLMs
- Provide text and image-based learning inside one seamless interface
- Keep the experience friendly, simple, and highly student-centric
They also needed a scalable architecture that could handle:
- Multi-modal content
- Multi-LLM orchestration
- Dynamic context management
- Specialised AI agents working independently yet in sync
Our Approach
We began with a 2–3 week research phase to understand the functional and cost behaviour of major AI systems. Our goal: engineer a setup that delivers high capability without blowing up operational expenses.
We then designed a modular architecture with dedicated AI agents for:
- Key Points
- Study Plan
- Diagrams & Diagram Generation
- Formulas
- Practice Tests
- General Chat
Each module had its own context rules, workflows, and LLM selection strategy, enabling the system to behave like a real mentor rather than a generic chatbot.
Solution Delivered
- Personalized Study Guidance
The AI creates tailored study plans based on the student’s course and subject. Plans are structured, achievable, and exam-oriented. - Keypoint Generator
Students receive crisp, exam-focused summaries for any topic or chapter, generated instantly. - Diagram Intelligence
The platform can both:- Explain textbook diagrams, and
- Generate new diagrams using image generation models, then combine everything into one fluid response.
- Formula Assistance
Students can ask any formula-related query and get explanations, breakdowns and usage examples. - Practice Test Engine
The system delivers topic wise practice questions along with:- Answer explanations
- Performance evaluations
- Difficulty-based question sets
- Smart Follow Up Suggestions
To make the experience engaging, the AI suggests the next logical questions a student may ask, increasing learning depth and retention. - Multi LLM Orchestration
We integrated multiple AI providers into a single pipeline, selecting the most efficient model per module to balance:- Accuracy
- Speed
- Cost
- Clean, Scalable Architecture
A hybrid storage system using MySQL + MongoDB, with Laravel & Node orchestrating business logic and agent coordination.
Impact
For Students
- Faster learning and clearer understanding
- Personalized support that feels like a real mentor
- All study tools — notes, diagrams, formulas, tests — in one unified space
For the Client
- A cost-optimized AI platform running multiple modules smoothly
- Differentiation from single-model competitor apps
- Scalable architecture ready for thousands of students
Technologies Used
- Backend: Laravel, Node.js
- Frontend: Flutter (mobile app)
- Databases: MySQL, MongoDB
- AI Providers: Multiple LLMs for specialized modules
- Image Models: Multi-provider (Serp API) diagram generation