README.md

---
title: CrewAI Multiagent Research Tool
emoji: 📈
colorFrom: blue
colorTo: blue
sdk: gradio
sdk_version: 5.9.1
app_file: app.py
pinned: true
license: cc-by-nc-nd-4.0
---
# CrewAI Multiagent Research Tool

Welcome to the **CrewAI Research Tool** on Hugging Face Spaces! This tool leverages **Gradio**, **Cohere**, **CrewAI**, and **Azure/OpenAI technologies** to automate and streamline online research, compile reports, and extract data through web crawling.

## 📝 Description

The CrewAI Research Tool uses advanced AI capabilities to:
- Gather, analyze, and compile information on various topics.
- Deliver comprehensive, structured reports.
- Automate the research process, from web searching to report generation.

## 🚀 Features

- **Web Search Automation**: 
  - Uses the DuckDuckGo search engine to perform searches and gather detailed results.
- **Data Extraction and Summarization**:
  - Extracts and summarizes web content using web scraping tools and LLMs.
- **Detailed Report Generation**: 
  - Agents like 'Researcher' and 'Editor' sequentially craft polished reports based on findings.
- **Interactive Gradio Interface**: 
  - Allows users to input research topics and generate reports with a single button click.

## 🛠 Tools and Technologies

- **CrewAI**: 
  - Orchestrates agents with specific roles to automate workflows.
- **OpenAI, Azure OpenAI Services, and Groq**: 
  - Employs large language models for understanding and content generation.
- **Gradio**: 
  - Provides an intuitive web interface for user interaction.

## ⚙️ Utility Functions

- **`setup_logging()`**: Configures logging to output messages to the console.
- **`export_to_markdown(result)`**: Saves the generated report to a Markdown file.
- **`fetch_content(url)`**: Retrieves content from a URL, handling both PDFs and web articles.

### Custom Tools

- **`search_results(search_query: str) -> list`**: 
  - Performs a web search using DuckDuckGo and returns a list of results.
- **`web_scrapper(url: str, topic: str) -> str`**: 
  - Extracts and summarizes content from a URL related to a topic.

## 🤖 AI Agents and Tasks

### Agents
1. **Researcher**: Gathers detailed information on a specified topic using search and scraping tools.
2. **Editor**: Refines the collected data into a structured, polished report.

### Tasks
- **`research_task`**: Directs the Researcher to collect and draft information.
- **`edit_task`**: Guides the Editor to refine drafts into final reports.

## 🧠 Crew Orchestration

The **Crew** class manages agents and tasks, ensuring proper execution order and seamless interactions.

## 💻 How to Use

The user interface, built with **Gradio**, includes:
- **Model Selection**: Choose between AI models (Azure, OpenAI, Groq-Cohere).
- **API Key Inputs**: Provide necessary API keys for the selected model.
- **Topic Input**: Enter the research topic.
- **Buttons**: 
  - **Start Research**: Initiates the research process.
  - **Export to Markdown**: Saves the report as a Markdown file.

### Steps
1. **Select a Model**: Use the radio buttons to choose your preferred AI model.
2. **Enter API Keys**: Fill in the required API keys based on the model.
3. **Enter a Topic**: Type your research topic into the textbox.
4. **Start Research**: Click the button to generate the report.
5. **View Results**: See the generated report in the output section.
6. **Export Report**: Save the report as a Markdown file for future reference.

## 🌟 Contributions and Feedback

We welcome contributions! 
- Duplicate this Space, make improvements, and submit a pull request.
- For feedback and issues, open an issue in the project's Community tab.

## 📜 License

This project is licensed under the **cc-by-nc-nd-4.0**.

## 🙏 Acknowledgments

Special thanks to the open-source projects and APIs that made this tool possible:
- [CrewAI](https://crewai.com)
- [OpenAI](https://openai.com)
- [Azure OpenAI Services](https://azure.microsoft.com/en-us/products/ai-services/openai-service)
- [Cohere](https://cohere.com)
- [Groq](https://groq.com)
- [DuckDuckGoSeach](https://pypi.org/project/duckduckgo-search/)

And of course to [Hugging Face](https://huggingface.co)!! 🤗