Report Generator Agent

# AI Research Report Agent

The AI Research Report Agent is an advanced, multi-agent system designed to automate the end-to-end research and report-writing workflow. Its primary function is to transform a user-provided topic into a comprehensive, well-structured, and polished report. The system's functionality is centered around a collaborative, approval-based process. 

### Key Features
- **Style Analysis:** Analyzes a user-provided document to replicate its writing style.
- **Outline Generation:** Creates a detailed report outline requiring user approval before proceeding (Human-in-the-Loop).
- **Iterative Content Creation:** Produces content section-by-section based on real-time research.
- **Bibliography Management:** Automatically formats citations in APA, MLA, or Chicago style.
- **Post-Generation Services:** Translates the final report into other languages and converts it into a professional, presentation-ready PDF document.

---

## Technologies Used

The project is built on a modern Python stack and leverages the Google Cloud ecosystem for its core AI and deployment capabilities. 

### Core Technology Stack
- **Framework:** Google Agent Development Kit (ADK).
- **Programming Language:** Python 3.11+ with pip for dependency management.
- **Generative Model:** Google’s Gemini 2.5 Pro model, accessed via the Vertex AI API.
- **Deployment Platform:** Google Cloud Run for scalable microservice deployment.

### Supporting Libraries
- **PDF Generation:** FPDF2 for creating PDF documents.
- **Content Conversion:** Markdown2 for converting content formats.

---

## Data Sources

The agent gathers information exclusively from live web searches to ensure the use of the most current and relevant information. 

- **Primary Data Source:** Real-time web data accessed via the integrated Google Search tool.
- **Component Responsible:** WebSearchAgent, which collects URLs of all sources used.
- **Bibliography:** All cited sources are included in the final report in the requested citation style.

---

## Findings and Learnings

### Key Insights
1. **Multi-Agent Architecture:** 
   - Using specialized agents (e.g., for style analysis, outline creation, writing) significantly improved reliability and quality compared to a monolithic approach.
2. **Human-in-the-Loop (HITL):**
   - Gaining explicit user approval at critical checkpoints ensures alignment with the user's vision and prevents wasted computation.
3. **Effective Orchestration:**
   - The Coordinator agent, managing workflow and state between agents, proved to be the most critical component for success.

This project highlights the effectiveness of combining multi-agent systems, real-time data, and collaborative workflows to create high-quality, tailored outputs.

Built With

• adk
• gemini
• google
• python