ResearchAgent

An AI-powered multi-agent research system that autonomously searches the web, extracts relevant information, generates structured research reports, and critiques its own output.

Built using LangChain, Groq LLMs, Tavily Search, and Streamlit.

Overview

ResearchAgent demonstrates how multiple AI agents can collaborate to solve a complex task by breaking it into specialized stages.

Instead of relying on a single prompt, the system uses a multi-agent workflow where each agent is responsible for a specific task:

1. Search Agent → Finds relevant information from the web.
2. Reader Agent → Extracts and processes content from selected sources.
3. Writer Agent → Produces a structured research report.
4. Critic Agent → Reviews the report and provides feedback.

This project was built to demonstrate practical AI Engineering concepts including:

• Tool Calling
• Agentic Workflows
• Multi-Agent Systems
• LLM Orchestration
• Prompt Engineering
• Web Search Integration
• Automated Research Pipelines

---

Architecture

User Query
    │
    ▼
┌─────────────────┐
│  Search Agent   │
│  Tavily Search  │
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│  Reader Agent   │
│ URL Extraction  │
│ + Web Scraping  │
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│  Writer Agent   │
│ Report Creation │
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│  Critic Agent   │
│ Quality Review  │
└────────┬────────┘
         │
         ▼
   Final Research
       Report

---

Features

• Multi-agent research workflow
• Web search using Tavily
• Automated content extraction
• Structured report generation
• Self-critique and evaluation
• Streamlit-based user interface
• Groq-powered LLM inference
• Modular architecture for easy extension

---

Tech Stack

AI & LLM

• LangChain
• LangGraph Runtime
• Groq API
• Llama 3.3 70B Versatile

Search & Retrieval

• Tavily Search API

Data Extraction

• BeautifulSoup
• Requests

Frontend

• Streamlit

Utilities

• Python
• dotenv
• Rich

---

Agent Design

Search Agent

Responsible for discovering relevant and recent information.

Tool Used:

web_search()

Responsibilities:

• Search the internet
• Collect relevant URLs
• Gather snippets and metadata
• Provide research context

---

Reader Agent

Responsible for retrieving deeper information from selected sources.

Tool Used:

scrape_url()

Responsibilities:

• Select useful resources
• Extract page content
• Clean HTML noise
• Prepare information for report generation

---

Writer Agent

Responsible for synthesizing information into a readable report.

Responsibilities:

• Combine gathered research
• Generate structured output
• Produce coherent findings
• Maintain factual consistency

Output Structure:

• Introduction
• Key Findings
• Conclusion
• Sources

---

Critic Agent

Responsible for evaluating report quality.

Responsibilities:

• Assess report completeness
• Identify weaknesses
• Suggest improvements
• Assign a quality score

Output Format:

Score: X/10

Strengths:
- ...

Areas to Improve:
- ...

Verdict:
...

---

Project Structure

ResearchAgent/
│
├── agents.py
├── app.py
├── pipeline.py
├── tools.py
│
├── requirements.txt
├── .env.example
├── .gitignore
│
└── README.md

File Responsibilities

File	Purpose
agents.py	Agent and chain definitions
tools.py	Search and scraping tools
pipeline.py	Multi-agent orchestration
app.py	Streamlit frontend
.env.example	Environment variables template

---

Screenshots

Home Page

Generated Research Report

Critic Feedback

Installation

Clone Repository

git clone https://github.com/yourusername/ResearchAgent.git

cd ResearchAgent

Create Virtual Environment

Windows

python -m venv venv

venv\Scripts\activate

Linux / Mac

python3 -m venv venv

source venv/bin/activate

Install Dependencies

pip install -r requirements.txt

---

Environment Variables

Create a .env file.

GROQ_API_KEY=your_groq_api_key
TAVILY_API_KEY=your_tavily_api_key

---

Running the Application

Streamlit UI

streamlit run app.py

Application will be available at:

http://localhost:8501

CLI Pipeline

python pipeline.py

---

Example Workflow

Input:

Gen AI Skills Required in 2026

Pipeline Execution:

Search Agent
   ↓
Reader Agent
   ↓
Writer Agent
   ↓
Critic Agent

Output:

Research Report
+
Quality Evaluation

---

Engineering Highlights

This project demonstrates:

• Multi-agent system design
• LLM tool calling
• Prompt orchestration
• Research automation
• Retrieval and synthesis workflows
• Modular agent architecture
• AI application development
• End-to-end GenAI system implementation

---

Future Improvements

• Async agent execution
• Multi-source scraping
• Citation tracking
• RAG-based knowledge memory
• Human-in-the-loop review
• Report export to PDF
• Agent observability and tracing
• Multi-model routing

---

Why This Project

Most LLM applications are simple chat interfaces.

This project focuses on AI Engineering principles by combining multiple specialized agents, external tools, autonomous decision-making, and structured report generation into a complete research workflow.

The goal is to demonstrate practical experience building agentic AI systems rather than simple prompt wrappers.

---

License

This project is licensed under the MIT License.