Research Assistant using Spring AI

This Research Assistant is a minimal, AI-powered research assistant designed to perform two high-value tasks: summarization and intelligent suggestion. It allows users to input raw text or queries and receive concise summaries along with relevant, context-aware suggestions to guide further exploration.

📦 Technologies

• Spring AI
• Gemini API
• Vite
• Java (Spring Boot)
• Javascript
• Tailwind CSS

🎯 Features

ResearchAssist focuses on doing a few things exceptionally well:

• Summarization Engine: Transform long-form content into concise, structured summaries without losing key meaning

• Contextual Suggestions: Generate relevant next steps such as related topics, follow-up questions, or areas to explore

👨‍🎓 The Process

This project began as a simple idea: "What if healthcare felt more human?" I spent many late nights sketching wireframes and imagining how technology could connect patients and doctors.

The journey started with research; I spoke with doctors, patients, and healthcare workers to grasp their real challenges. Then I faced technical hurdles: creating a voice interface that feels natural, developing a backend that securely manages sensitive medical data, and designing a mobile app that operates as smoothly as the web version.

🏗️ How I Built It

Building WellNest was like conducting an orchestra; every piece needed to work in harmony:

Phase 1: Backend Foundation (Spring Boot)

A lightweight and secure backend handles request processing and rate limiting. The architecture is modular, ensuring AI-related logic remains isolated and extensible.

Phase 2: AI Integration (Spring AI + Gemini API)

This is the core functional layer.

Summarization Flow:

• Input text is processed through structured prompts
• The Gemini API generates a concise, context-preserving summary

Suggestion Flow:

• The same input (or its summary) is analyzed
• The system generates:
  • Related topics
  • Follow-up questions
  • Suggested directions for deeper research

Spring AI orchestrates prompt management and response handling, ensuring consistency and clarity.

Phase 3: Frontend (Vite)

The frontend is intentionally minimal:

• Input field for text or query
• Output sections for:
  • Summary
  • Suggestions

Built with Vite for speed and responsiveness, the UI prioritizes clarity over feature bloat.

Phase 4: Deployment

• Frontend deployed on Vercel/Netlify
• Backend deployed as a Spring Boot service
• Gemini API securely integrated via environment variables

🏫 What I Learned Along the Way

This project taught me many important lessons:

• Spring Boot (Java): Narrowing the scope to summarization and suggestion enabled a clean, modular backend architecture, improving maintainability and overall usability
• Spring AI + Gemini API: Leveraging Spring AI for prompt orchestration with the Gemini API emphasized how precise prompt engineering directly affects summary quality and suggestion relevance
• Spring Boot (Service Layer Architecture): Separating summarization and suggestion into distinct service layers improved scalability, debugging, and extensibility
• Spring Boot + Vite (Performance Optimization): Optimizing end-to-end latency across the Spring Boot backend and Vite frontend ensured fast, responsive interactions essential for AI-driven workflows

Overall Growth:

Each part of this project helped me understand more about building apps, managing complex information, and improving user experience. It was more than just making a tool. It was about solving problems, learning new things, and improving my skills for future work.

🔧 How It Could Be Improved

• Streaming Responses: Real-time summary generation
• Custom Summary Modes: Bullet points, abstract, or detailed breakdowns
• Domain-Specific Suggestions: Tailored outputs for technical, academic, or general content
• File Input Support: Upload PDFs or documents directly
• History Tracking: Save and revisit previous summaries

🪧 Running the Project

To run the project in your local environment, follow these steps:

1. Clone the repository
2. Start backend:

   ./mvnw spring-boot:run

3. Start frontend:

   npm install
   npm run dev

4. Open http://localhost:5173

🎥 Video

research.assistant.mp4

📍 Preview

Visit website