🚀 GraphLens AI

GraphLens AI is an AI-powered research workspace. It integrates document-based vector search (RAG), real-time web search, and an autonomous file management assistant into a secure, user-isolated chat platform.

---

📸 Demo / Preview

---

🔄 System Flow

1. Ingestion: The user uploads a PDF, Word, or Text document. The Django backend parses and extracts text using PyMuPDF or python-docx.
2. Indexing: Extracted content is chunked, embedded, and indexed into Pinecone Serverless, tagged with the user's specific ID to ensure multi-tenant isolation.
3. User Query: The user sends a prompt through the chat interface.
4. Supervisory Routing: The LangGraph Supervisor inspects the query to check if it's a file operation (e.g. creating/reading files), or dynamically routes it to the correct worker agent.
5. Agent Execution:
   • Document Agent: Performs a similarity search in Pinecone scoped to the user's active document.
   • Research Agent: Conducts a live internet search on DuckDuckGo and deep-scrapes URLs using Jina Reader.
   • Workspace File Agent: Creates, reads, updates, or deletes files directly inside the user's directory on the server.
6. Synthesis & Sync: The consolidated context is passed to the Groq LLM to generate a grounded response. Any files created or modified by the File Agent are automatically re-indexed back into Pinecone.

---

🚀 Key Features

🧠 Intelligent Agent Routing (LangGraph)

• Built using LangGraph Supervisor Architecture.
• Intent-Based Logic: Decides whether a query requires document Q&A, a live web search, or workspace file operations.

🔍 Isolated Document RAG

• Multi-Tenant Privacy: Scopes document ingestion and vector database retrieval by user session to keep files completely private.
• Semantic Vector Search: Powered by Pinecone to query text chunks using cosine similarity.

🌐 Real-Time Web Research

• Live Search: Fetches real-time internet facts using the DuckDuckGo Search API.
• URL Deep-Scraping: Converts links into clean markdown content via the Jina Reader API for the LLM to analyze.

📂 Chat-Based File Manager

• Workspace CRUD: Allows users to manage server files (create, read, append, overwrite, delete) directly through chat instructions.
• Auto-Sync Indexing: Automatically indexes file edits and creations in the background database.

---

🎯 Use Cases

• 📚 Academic Research: Ask questions about reference PDFs and instantly verify claims on the live web.
• 📄 Workspace Note-Taking: Summarize documents or web research and save the results as new files in your workspace.
• 🔒 Isolated Data Auditing: Securely compare local records against current web search results.

---

🛠️ Tech Stack

• Frontend: Vanilla HTML5, CSS3, and JavaScript (served directly via Django templates).
• Backend: Django & Django REST Framework (DRF).
• AI Orchestration: LangChain & LangGraph.
• AI Engine: Groq Cloud API (llama-3.1-8b-instant).
• Vector Database: Pinecone Serverless (with llama-text-embed-v2 embeddings).
• Relational Database: Neon PostgreSQL (production) or SQLite (local development).
• Cloud Storage: AWS S3 (integrated via boto3 using presigned URLs).

---

🚀 Future Roadmap

• 🔹 Google Workspace Integration (Docs, Drive, Gmail)
• 🔹 Multi-File Chat: Simultaneous querying across multiple documents.
• 🔹 Persistent Memory: Longer-term session history for returning users.
• 🔹 Enterprise Auth: Production-grade user authentication systems.

---

📬 Contact

📧 vinaygattu005@gmail.com
🔗 GitHub: Vinay50029

---

⭐ If you like this project, give it a star on GitHub — it helps a lot!