Agent Research

Ever felt lost trying to keep up with all the new research and tech advancements?
Agent Research is here to make that easier.

Using advanced LLM capabilities, web search integration, and a multi-agent architecture, Agent Research can:

1. Search the web for up to 10 research papers on any topic, with filtering options like recency and relevance.
2. Summarize complex, jargon-heavy research papers into simple language.
   Don’t feel like reading? You can even listen to the summary like a podcast.
3. Analyze any research paper from just its PDF URL or DOI — showing you the summary, authors, and publish date.
4. Use multi-agent synthesis to combine insights from multiple papers and generate a unified explanation.
   For example, it can answer: "Why do Transformers need attention?" based on collective research.

---

⚙️ Tech Used

• Streamlit – for building the interactive UI
• Gemini API (Google Generative AI) – for summarization, classification, and synthesis
• PyMuPDF – for PDF text extraction
• arXiv API – for research paper discovery
• CrossRef API – for DOI metadata retrieval
• gTTS (Google Text-to-Speech) – to generate audio summaries
• Requests – to fetch content via URLs

---

🚀 Want to Get Started?

Try the hosted app directly here

---

🛠️ Want to Contribute / Run Locally?

1. Clone or fork the repo
2. Create a virtual environment and install dependencies using pip install -r requirements.txt
3. Create a .env.local file and add this: GEMINI_API_KEY=your_key_here
4. Run the app: streamlit run app.py

🎥 Working Demo

Research.Agent.mp4

🧩 Limitations and Future Scope

1. User Authentication & Personalization:
   Currently, the app does not store user data. A future version could include authentication and user-specific storage to retain their search history, saved summaries, and preferences.

2. Enhanced Audio Podcast Features:
   The current audio is generated using a basic TTS engine. Future improvements could include multi-language support, custom voices, tone variation, and podcast-style narration.

3. Smarter DOI Processing:
   While basic DOI-based extraction works, a more advanced system could directly fetch and process full research papers from various publishers and repositories using just the DOI, even in restricted access scenarios.