Multi‑Agent Automated Research & Report Generation

Overview

This repository implements a multi‑agent system for automated research and report generation using modern LLM tooling and structured orchestration. The system accepts a research topic from the user, autonomously gathers information from the web, analyzes and synthesizes findings, and produces a fully formatted research report (PDF or DOCX).

The architecture follows an agent‑oriented design and uses LangGraph to orchestrate execution order, state transitions, and retries.

---

Use Case

Automated Research & Report Generation

Goal

Produce a comprehensive, structured research report on a given topic using multiple specialized agents and integrated tools.

---

System Architecture

Primary Agents

1. Research Agent

   • Searches the web for relevant sources
   • Collects raw content, snippets, and metadata
   • Returns structured research results

2. Analysis Agent

   • De‑duplicates and ranks sources by relevance
   • Extracts key insights and themes
   • Produces a structured outline / analytical brief

3. Report Writer Agent

   • Converts the structured brief into a complete report
   • Applies formatting and sectioning
   • Exports the final document to PDF or DOCX

---

Tools Integrated

• Web Search Tool Wrapper around SerpAPI / Bing / Google Custom Search. If API keys are not configured, a mock search provider returns curated demo results for local development.

• Summarization / Extraction Tool Lightweight summarization utilities powered by Google GenAI, optimized for condensing long research content.

• Document Generator Tool Generates professional documents using Python libraries such as python-docx, weasyprint, or reportlab.

---

Orchestration (LangGraph)

LangGraph manages the entire execution pipeline as a directed workflow graph:

start
↓
research_node
↓
analyze_node
↓
write_node
↓
export_node

Responsibilities of LangGraph:

• Enforces execution order
• Manages shared state between agents
• Handles retries and failure recovery
• Enables future branching or parallelism

---

Workflow

1. User submits a research topic via the UI.
2. Research Agent invokes the Web Search Tool to gather sources.
3. Analysis Agent synthesizes and structures the research output.
4. Report Writer Agent generates a formatted document.
5. Document Generator exports the final report (PDF / DOCX).

---

Summary Table

Use Case	Purpose	Agents Involved	Tool Integration	Orchestration Role
Automated Research & Report Generation	Produce a fully generated research document	Research, Analysis, Report Writer	Web search, summarization, document generator	Sequential pipeline managed by LangGraph

---

High‑Level Design

Agents

• ResearchAgent Gathers web content using a Web Search tool (SerpAPI optional) and stores raw snippets with metadata.

• AnalysisAgent Ingests research results, removes duplicates, ranks relevance, extracts outlines, and produces a structured analytical brief.

• ReportWriterAgent Converts the structured brief into a formatted Markdown or HTML report and exports it to PDF or DOCX.

Tools

• WebSearchTool Abstracts external search APIs. Falls back to a mock provider for offline development.

• SummarizationTool Uses Google GenAI to summarize and extract key points from long text segments.

• DocumentGeneratorTool Responsible for rendering and exporting documents using Python document libraries.

LLM

• Powered by Google GenAI (Gemini) via the official Python SDK.
• Includes a safe mock LLM client for local testing and CI environments.

---

User Interface

• Gradio Web UI

  • Topic input field
  • Configuration options (depth, number of sources)
  • Run pipeline button
  • Download generated report

---

Repository Structure

multiagent-research-report/
├─ README.md
├─ requirements.txt
├─ .env.example
├─ start.sh
├─ Dockerfile                  # optional
├─ app.py                      # Gradio UI + orchestration entry point
├─ src/
│  ├─ config.py                # central configuration loader
│  ├─ orchestrator/
│  │  └─ langgraph_workflow.py # LangGraph workflow definition
│  ├─ agents/
│  │  ├─ research_agent.py
│  │  ├─ analysis_agent.py
│  │  └─ report_writer_agent.py
│  ├─ tools/
│  │  ├─ web_search.py         # SerpAPI / mock search wrapper
│  │  ├─ summarizer.py         # summarization utilities
│  │  └─ doc_generator.py      # PDF / DOCX generator
│  └─ utils/
│     └─ llm_client.py         # Google GenAI client wrapper
└─ examples/
   └─ sample_report.pdf

---

Setup & Installation

git clone https://github.com/abdi7d/multiagent-research-report.git
cd multiagent-research-report
pip install -r requirements.txt
cp .env.example .env

Configure API keys in .env if using live web search and LLMs.

---

Running the Application

python app.py

The Gradio interface will launch in your browser.

---

Future Enhancements

• Parallel research agents
• Citation tracking and reference management
• Long‑term memory with vector databases
• Multi‑language report generation
• Academic citation formats (APA, IEEE)

---

License

MIT License

---

Acknowledgements

• LangGraph
• Google GenAI (Gemini)
• Gradio
• Open‑source Python ecosystem