Inspiration

Students often struggle to convert raw notes into structured, exam-ready material. Long paragraphs, unorganized content, and complex explanations make revision difficult and time-consuming.

At the same time, many people search online about health symptoms and receive alarming or misleading information. We wanted to build a system that not only improves academic understanding but also promotes responsible AI use in sensitive areas like health.

This inspired us to create an AI assistant that focuses on clarity, structure, and ethical guidance.

What it does

Our project is an AI-powered assistant that performs two intelligent functions:

Academic Transformation Mode

Converts raw notes into structured, point-wise explanations

Generates beginner-friendly explanations

Creates 5-mark and 10-mark exam answers

Extracts key terms and definitions

Provides quick revision summaries

Responsible Health Guidance Mode

Automatically detects symptom-related input

Does not diagnose or mention diseases

Responds calmly and responsibly

Suggests general care steps

Encourages consulting a healthcare professional

The system switches behavior silently based on input context.

How we built it

We built the project using:

Flask (Python) for backend

Gemini 3 Flash model for AI reasoning

Structured system prompts to control behavior

Keyword-based health detection logic

Context-aware response switching

A clean HTML/CSS interface for user interaction

We designed conditional logic to differentiate between academic content and health-related inputs.

Challenges we ran into

Preventing false health-mode triggers

Designing non-diagnostic health responses

Maintaining structured academic output consistently

Avoiding long, unstructured AI responses

Balancing usefulness with ethical responsibility

We had to carefully engineer prompts and detection logic to ensure safe and structured outputs.

Accomplishments that we're proud of

Successfully built context-aware AI behavior

Implemented silent auto-detection of health-related input

Enforced structured academic formatting

Designed ethical safeguards for sensitive topics

Built a dual-purpose AI assistant with clear role control

We are proud of creating a system that is both intelligent and responsible.

What we learned

Prompt engineering significantly affects output quality

Structured formatting improves user understanding

Responsible AI design requires clear behavioral constraints

Context detection improves realism and usability

Ethical considerations are as important as technical performance

What's next for enhanced usage of Gemini 3

Smarter semantic detection instead of keyword-based health triggers

PDF and document upload for automatic academic conversion

Multi-language support

Voice input for accessibility

Adaptive learning modes (Beginner / Exam / Interview)

Image-based content understanding using Gemini’s multimodal capabilities

We aim to further leverage Gemini 3’s reasoning and multimodal strengths to make the system more intelligent and accessible

Built With

Share this project:

Updates