DocQuery 🧠📄

DocQuery is an AI-powered document question-answering app.
Upload a text-based PDF, TXT, or DOCX file and ask questions in natural language.
Answers are grounded strictly in the uploaded document using vector search + OpenAI models.

Built for:

• Fast indexing
• Clear, source-based answers
• Clean chat UX
• Safe credit usage
• One-document-at-a-time workflow

---

✨ Features

• 📄 Supports PDF, TXT, DOCX
• 🚫 Rejects scanned/image-only PDFs
• ⚡ Fast bounded indexing (usually under 1 minute)
• 🔍 Semantic search using FAISS + OpenAI embeddings
• 💬 Natural chat interface
• 🧠 Answers strictly from document content (no hallucinations)
• 🧭 Follow-up suggestion buttons when information isn’t found
• 🔒 UI lock while indexing or answering
• ♻️ Replace or remove active documents anytime
• 🧹 Clear chat button

---

🛠️ Tech Stack

• Frontend: Streamlit
• Vector Store: FAISS
• Embeddings: OpenAI text-embedding-3-small
• LLM: OpenAI gpt-4o-mini
• Framework: LangChain

---

🚀 Run Locally

1. Clone the repo:

git clone https://github.com/<your-username>/docquery.git
cd docquery

🔑 OpenAI API Key Required

This app does not ship with an API key.
You must use your own OpenAI API key to run DocQuery.

Your key is:

• Never stored in this repository
• Never committed to GitHub
• Loaded from environment variables or Streamlit Secrets

Local usage:

Create a .env file:

OPENAI_API_KEY=sk-your-own-key-here