Guru Intelligence

Knowledge at your fingertips

What it does

This application is a document analysis and question-answering system built on Next.js, using Google's Gemini API and Pinecone vector database. It implements Retrieval Augmented Generation (RAG) to provide intelligent responses based on uploaded PDF documents.
Users can upload PDFs, which are processed and stored as embeddings in Pinecone using Gemini Embeddings and LangChain. This enables efficient retrieval of relevant information for user queries. It also provides excellent User Interface.
The app features five modes:

MCQ: Generates multiple-choice questions
Explain: Provides detailed explanations
Summarize: Creates concise summaries
Bullet Points: Extracts key points
Table Comparison: Compares information in tabular format

When a user selects a mode and submits a query, the system retrieves relevant document sections from Pinecone and uses Gemini 1.5 Flash API to generate tailored responses.
This RAG approach combines pre-trained language models with real-time access to document content, ensuring context-aware and accurate answers. The use of Gemini API for both embeddings and response generation provides high-quality results across various query types and analysis tasks.

Built with

  • Web/Chrome
  • Gemini API

Team

By

Ayush, Nachiketh, Rishabh

From

India