🚀 OpenDeepResearcher

OpenDeepResearcher is an AI-powered multi-agent research assistant that autonomously plans, searches, analyzes, and synthesizes information into structured reports. The system uses locally hosted large language models via LM Studio, enabling private, cost-efficient, and intelligent deep research workflows without relying on cloud LLM APIs.

---

✨ Features

• 🔍 Autonomous research planning with subquestion generation
• 🌐 Real-time web search using Tavily API
• 🧠 Multi-agent reasoning pipeline (Planner → Searcher → Writer)
• 📝 Academic-style report generation with citations
• 🔒 Local LLM inference using LM Studio (privacy-friendly)
• 📊 Structured JSON research plans
• ⚡ Modular and extensible architecture

---

🏗️ System Architecture

The system follows an agent-based pipeline:

1. Planner Agent

   • Breaks user queries into 6–8 focused subquestions
   • Generates structured research plans
   • Extracts constraints and keywords

2. Searcher Agent

   • Retrieves relevant sources using Tavily API
   • Collects titles, URLs, snippets, and relevance scores

3. Writer Agent

   • Synthesizes findings into a comprehensive academic report
   • Adds citations and references
   • Saves final output to file

---

🤖 Model & Runtime

LLM Runtime: LM Studio
Model: Qwen 2.5 7B Instruct
Inference: Local (offline capable)

Benefits:

• No cloud dependency
• Improved privacy
• Lower cost
• Full control over model execution

---

📂 Project Structure

OpenDeepResearcher/
│── agents/
│     ├── planner_agent.py
│     ├── searcher_agent.py
│     └── writer_agent.py
│
│── main.py
│── requirements.txt
│── README.md
│── LICENSE
│── .gitignore

---

⚙️ Installation

Clone the repository:

git clone https://github.com/diamehak/OpenDeepResearcher.git
cd OpenDeepResearcher

Create virtual environment:

python -m venv venv
source venv/bin/activate   # Mac/Linux
venv\Scripts\activate      # Windows

Install dependencies:

pip install -r requirements.txt

---

🧠 LM Studio Setup

1. Install LM Studio
2. Download Qwen 2.5 7B Instruct model
3. Start the LM Studio local server
4. Ensure API endpoint is running:

http://127.0.0.1:1234

---

🔑 API Key Configuration

This project uses the Tavily Search API for retrieving research sources.

Before running the project, open the file:

agents/searcher_agent.py

Locate the API key placeholder:

ENTER_YOUR_TAVILY_API_KEY_HERE

Replace it with your actual Tavily API key.

You can obtain a Tavily API key from: https://tavily.com

▶️ Usage

Run the main program:

python main.py

Enter your research question when prompted.

Example:

Impact of artificial intelligence on healthcare diagnostics

The system will:

1. Generate subquestions
2. Search for sources
3. Produce a structured research report
4. Save output to a file

---

📊 Example Output

The system generates:

• Research plan JSON file
• Detailed terminal logs
• Final academic report (.txt)

---

🎯 Use Cases

• Academic research assistance
• Literature reviews
• Market analysis
• Technical research
• Policy research
• Automated report generation

---

🔮 Future Improvements

• Memory-enabled agents
• Multi-modal document ingestion (PDFs)
• Citation formatting (APA/IEEE)
• Web UI dashboard
• Vector database integration
• Streaming responses

---

🛠️ Tech Stack

• Python
• LM Studio
• Qwen 2.5 7B
• Tavily Search API
• Requests
• Agentic AI architecture

---

🤝 Contributing

Contributions are welcome!

1. Fork the repository
2. Create a feature branch
3. Commit changes
4. Open a Pull Request

---

📜 License

This project is licensed under the MIT License.

---

👤 Author

Dia Mehak AI / ML Developer | Agentic AI Enthusiast

---

⭐ If you find this project useful, please consider giving it a star!