AI-Generated Quarterly Financial Reports at Scale
Ricardo Ishida
Project Overview
This project was designed to automate the entire process of producing quarterly financial reports. Traditionally time-consuming and dependent on multiple analysts, this solution allows users to simply input a company name — and within minutes, receive a full 30-slide strategic report with charts, insights, and benchmarks against top competitors. The system combines real-time data extraction, multi-agent analysis, and AI-powered storytelling to deliver polished presentations ready for decision-makers, consultants, or investors.
Solution Design
• FMP API – Real-time financial data (income statement, balance sheet, key ratios, etc.)
• CrewAI – Multi-agent framework to orchestrate tasks like metric analysis, peer benchmarking, and insight generation
• OpenAI GPT models – Narrative generation and slide text creation • Matplotlib / Seaborn – Automated chart generation
• python-pptx (Python library) – Slide creation engine
• LlamaIndex + RAG (optional) – Enhanced answers based on context from filings, earnings calls, and reports
• AWS S3 or custom server – For storing and sharing reports automatically
Workflow Breakdown
1. User Input: Company name is entered via a simple interface
2. Data Collection: System queries the FMP API for:
• Latest quarterly financials
• Peer group identification and competitor data
3. Analysis Phase (CrewAI): Multiple AI agents are activated to:
• Benchmark company vs. industry
• Detect performance trends (revenue, EBITDA, margins)
• Highlight risks, opportunities, and anomalies
4. Report Generation:
• Charts are created automatically from the processed data
• Slide titles and narratives are written using OpenAI
• Slide text is highlighted (bold, italic, underlined) and sources are linked
• A clean, branded 30-slide PowerPoint is compiled using python-pptx
5. Delivery: The final report is:
• Saved locally or uploaded to an S3 bucket / cloud directory
• Optionally enhanced with contextual data via RAG (LlamaIndex) to train the model with your company's documentation
Setting Up the AI Agent using CrewAI
Key Innovations
• Multi-agent system: Modular architecture that allows each AI agent to specialize in one part of the report (financials, benchmarking, risk analysis, forecasting)
• Plug-and-play data layer: Works with any company supported by FMP API, enabling scalable reporting across industries
• Fully automated delivery: From input to polished deck in under 10 minutes
• Expandable RAG layer: Can incorporate external content like earnings transcripts, press releases, or investor letters for deeper insights
Like this project
Posted Apr 2, 2025
Built an end-to-end AI system that generates quarterly financial reports for any public company
Likes
0
Views
7
