DocAI

A Document Understanding AI system that gleans insights from PDF documents.

Features

• CLI Frontend: Provides an easy-to-use command-line interface for user interactions.
• Web Frontend: Provides a very simple interactive UI.
• Content Processing: Capable of accepting and processing information from PDF files.
• Natural Language Processing: Receives, understands, and provides feedback in natural language.
• Context-Aware: Understands conversation context and answers questions accordingly.
• Retrieval and Generation: Delivers answers grounded in facts, minimizing the chances of hallucination.
• Vector Storage: Stores and retrieves text vector representations.
• Feedback Mechanism: Allows users to provide additional context (e.g., more PDFs).

Tech Stack

• Python
• LangChain
• ChromaDB
• Groq Language Models
• Streamlit

Installation

• Clone this repository: git clone https://github.com/Adlai-1/DocAI.
• Install the needed packages: pip install -r requirements.txt (must be executed in the project directory)
• Add your API key to config.ini You can obtain an API key from GroqCloud (https://console.groq.com/keys)
• Execute "python -m streamlit run web/home.py" to run the WebUI.
• Execute "python ui/server.py" to start the server before running the User client using "python ui/client.py"

DocAI snapshots