PRISM is an AI-powered assistant designed for investors, analysts, and shareholders. It leverages Retrieval-Augmented Generation (RAG), Google Gemini, LangChain, and ChromaDB to make financial reports accessible, understandable, and actionable.
- π RAG-based Question Answering: Ask complex questions about lengthy financial reports and get precise, context-aware answers.
- π PDF Document Parsing: Upload annual reports in PDF format β the app automatically extracts and indexes content.
- π§ LangChain Integration: Orchestrates the entire pipeline of document loading, chunking, embedding, and querying.
- π€ Google Gemini LLM: Generates high-quality, insightful responses in natural language.
- ποΈ ChromaDB Vector Store: Efficiently stores and retrieves embedded document chunks for fast contextual answers.
LangChainfor chaining and orchestrationGoogle Geminifor LLM-powered reasoningChromaDBfor vector storage and retrievalPyMuPDFfor PDF parsingHuggingFace Embeddingsfor document chunk vectorization
- Clone the repo
-
git clone https://github.com/aymanhaque/HackAI25.git cd HackAI25 pip install -r requirements.txt python3 backend/new.py - On a different terminal:
-
cd HackAI25 cd frontend/my-app npm install npm run dev