🧠 Research Agent

An AI-powered assistant that helps you search for research papers, scrape academic content, summarize it using large language models (LLMs), extract and summarize PDF files, and even generate hypotheses — all in one easy-to-use Streamlit interface.

🚀 Features

• 🔍 Literature Search from arXiv via DuckDuckGo
• 🌐 Web Scraping of abstracts and page content
• 📄 PDF Reader with text extraction (via PyMuPDF)
• 🤖 Summarization & Hypothesis Generation using ibm-granite/granite-3.3-8b-instruct (via Replicate)
• 🧪 Ideal for students, researchers, and academics

---

📸 Demo

---

📦 Requirements

Install all dependencies using:

pip install -r requirements.txt

Or individually:

pip install streamlit replicate requests beautifulsoup4 PyMuPDF

---

🔑 Replicate API Setup

1. Go to Replicate API Tokens
2. Generate a new API token
3. Recommended: Set it as an environment variable

export REPLICATE_API_TOKEN=your_token_here

Or, you can hardcode it (not recommended for production):

REPLICATE_API_TOKEN = "your_token_here"
client = replicate.Client(api_token=REPLICATE_API_TOKEN)

---

🧠 Model Used

• Model: ibm-granite/granite-3.3-8b-instruct
• Version ID: 3ff9e6e20ff1f31263bf4f36c242bd9be1acb2025122daeefe2b06e883df0996

You can change the model in app.py if needed.

---

🖥️ Run the App

Start your Streamlit app locally:

streamlit run app.py

Then go to LocalHost.

---

✍️ Example Prompts

Find three recent research papers on the ethical implications of using CRISPR technology in humans.
Summarize the uploaded paper and highlight the main results and methodology.
Suggest a hypothesis based on the uploaded PDF.
Draft an abstract for a paper on AI in climate modeling.

---

📁 File Structure

📦 research-agent/
├── app.py              # Main Streamlit app
├── requirements.txt    # Required libraries
└── README.md           # You're here

---

🛡️ Disclaimer

• This tool uses third-party APIs and AI models. Use it responsibly for academic and research purposes only.
• Always verify LLM-generated content before citing or using it in official research.

---

💡 Future Improvements

• Export to PDF / BibTeX / CSV
• Follow-up Q&A on paper content
• Semantic search with vector embeddings

---

👤 Author

Samarth Pujari

AI Intern @ IBM

Connect with me on LinkedIn | Kaggle