DocuChat AI is the ultimate AI-powered assistant that lets you chat with your documents! 📄✨
It’s designed for students, researchers, and professionals who need instant insights from PDFs, DOCX, TXT, and CSV files.
✔ Multi-Format Uploads: Supports PDFs, Word Docs, TXTs & CSVs
✔ Smart AI Search: Uses DeepSeek R1 & GPT-4o for better answers
✔ Web Integration: If the doc lacks info, AI fetches real-time data from Google
✔ FAISS Vector Search: Efficient memory retrieval for fast answers
✔ Advanced Analytics: 🚀- Tracks peak usage days & AI sources (RAG, GPT, Web)
🔗 Try It Now → DocuChat AI
We now track every user query, storing response sources (RAG, GPT-4o, or Web).
This provides businesses & researchers with key insights like:
✅ Peak Usage Days – Know when users are most active 📊
✅ Most Used AI Model – RAG vs. GPT vs. Web 🔍
✅ User Engagement Trends – Optimize performance 💡
Why This Matters?
Businesses can now see where users get stuck, helping them improve knowledge bases, FAQs, and support docs. 🚀
This guide provides a detailed breakdown of the files in the repository, explaining their purpose and how they contribute to the app’s functionality.
| File | Purpose |
|---|---|
.env |
Stores API keys (GPT, DeepSeek, Web Search) 🔑 |
.gitignore |
Ensures sensitive files (API keys, cache) aren’t uploaded 🚫 |
main.py |
Core AI Logic – Handles uploads, AI processing, & responses 🤖 |
app.sh |
Auto Setup – Launches the app in one click 🚀 |
requirements.txt |
Dependency List – Install all required packages 📦 |
setup.sh |
Automates virtual environment & installation 🛠 |
✅ LangChain – Manages AI-driven document retrieval
✅ FAISS – Speeds up document searches
✅ GPT-4o & DeepSeek R1 – AI-powered text generation
✅ Streamlit – User-friendly UI
✅ SerpAPI / Bing API – Web search fallback
git clone git@github.com:cousintiz/Marta-GPT-v0.0.1.git
cd Marta-GPT-v0.0.1python3 -m venv venvsource venv/bin/activatevenv\Scripts\activatevenv\Scripts\Activate.ps1pip install -r requirements.txtstreamlit run main.pyOnce the app starts, it will automatically open in your default browser. If not, go to:
http://localhost:8501
- Create an OpenAI account if you don’t have one.
- Get your API key here
- Enter the API key in the app when prompted.
🔗 Try it live: DocuChat AI
✅ No setup needed
✅ Upload & chat instantly
✅ Works on any device
1️⃣ User Uploads a Document
- Supported formats: PDF, DOCX, TXT, CSV
- File is temporarily stored for processing
2️⃣ AI Reads & Understands
- Uses LangChain + FAISS for fast document retrieval
- Creates a vector store for better memory
3️⃣ AI Answers User Questions
- First, retrieves relevant info from the doc (RAG)
- If no answer is found, GPT-4o generates a response
- If GPT also doesn’t know, it searches Google/Bing
4️⃣ **Insights Are Logged **
- ✅ Timestamps are saved
- ✅ AI Source (RAG, GPT, Web) is recorded
- ✅ Data is analyzed to track trends
5️⃣ Secure & Clean
- AI deletes uploaded files after each session
- User queries are anonymized for privacy
✔ Saves Hours of Manual Searching
✔ Instant AI-Powered Insights
✔ Web Search When Docs Lack Data
✔ Enterprise-Grade Security
✔ Now With AI Usage Analytics!
🔗 Start Chatting Now → DocuChat AI
We welcome open-source contributions!
✅ Improve document parsing
✅ Add multi-language support
✅ Optimize search & retrieval
✅ Fix bugs & issues
✅ Suggest new features 🚀
This project is MIT Licensed, meaning it’s free to use, modify, and share.
1️⃣ Clone the repo
2️⃣ Install dependencies
3️⃣ Set up your API keys
4️⃣ Start chatting with your documents! 🚀
📌 Need Help? Contact us on GitHub Discussions! 😊