Financial Document Analyzer

AI-powered full-stack application that analyzes financial PDF documents and produces structured, evidence-based financial, investment, and risk insights.

Live Demo

• Frontend: https://financial-document-analyser.netlify.app/

Overview

• Frontend: React + Vite (client/)
• Backend: FastAPI + CrewAI + Groq LLM (server/)
• Input: financial PDF + user query
• Output: consolidated analysis report from a multi-agent workflow

Project Structure

client/   Frontend app and API client
server/   FastAPI API, CrewAI agents/tasks/tools
README.md Main documentation

Bugs Found And Fixes Applied

1. API routing mismatch between dev/prod

• Symptom: frontend calls failed depending on environment.
• Cause: API URL construction was not robust for proxy and deployed backends.
• Fix: improved client/src/services/api.js to:
  • use VITE_API_BASE_URL when provided,
  • use same-origin /analyze in Vite dev,
  • fallback to local backend for production preview.

2. CORS failures when frontend/backend ran on different hosts

• Symptom: browser blocked /analyze requests.
• Cause: strict origin handling.
• Fix: backend CORS now supports configured ALLOWED_ORIGINS and regex-based local origins.

3. CrewAI/Groq rate-limit failures

• Symptom: RateLimitError and unstable pipeline runs.
• Cause: multi-step calls and repeated document reads increased RPM/TPM pressure.
• Fix:
  • centralized crew limiter (max_rpm),
  • retry-after parsing + delayed retry,
  • compact PDF extraction with PDF_CHAR_LIMIT,
  • reduced redundant tool usage in downstream tasks.

4. File-processing instability (Stream ended unexpectedly)

• Symptom: intermittent backend errors after upload.
• Cause: aggressive file cleanup timing.
• Fix: removed immediate delete-on-finally behavior and stabilized processing flow.

5. UX quality issues during long runs

• Symptom: generic loading feedback and inconsistent form visuals.
• Fix:
  • process-aware loader with active agent + stage status,
  • elapsed timer during analysis,
  • consistent input styling for disabled/readonly state.

Setup Instructions

Prerequisites

• Python 3.10+
• Node.js 18+ (Node 20 recommended)
• Groq API key

1. Backend Setup

cd server
python -m venv .venv
.venv\Scripts\activate
pip install -r requirements.txt

Create server/.env from server/.env.example:

GROQ_API_KEY=your_groq_api_key
SERPER_API_KEY=
DEFAULT_QUERY=Analyze this financial document for investment insights
DATA_DIR=data
API_RELOAD=true
ALLOWED_ORIGINS=http://localhost:5173,http://127.0.0.1:5173
ALLOWED_ORIGIN_REGEX=^https?://(localhost|127\.0\.0\.1)(:\\d+)?$
PDF_CHAR_LIMIT=9000

Run backend:

uvicorn main:app --host 127.0.0.1 --port 8000 --reload

2. Frontend Setup

cd ..\client
npm install

Create client/.env from client/.env.example:

VITE_API_BASE_URL=
VITE_API_ANALYZE_PATH=/analyze

Run frontend:

npm run dev

Open http://localhost:5173.

Usage

1. Open the app in browser.
2. Upload a financial PDF.
3. Enter or keep the default analysis query.
4. Click the run button.
5. Monitor active agent and stage progress.
6. Review structured analysis output.

API Documentation

Base URL (local): http://127.0.0.1:8000

GET /

Health endpoint.

Response:

{"message":"Financial Document Analyzer API is running"}

POST /analyze

Analyzes an uploaded PDF using CrewAI workflow.

Request type:

• multipart/form-data

Form fields:

• file (required): PDF file
• query (optional): analysis instruction string

Success response (200):

{
  "status": "success",
  "query": "Analyze this financial document for investment insights",
  "analysis": "...agent output...",
  "file_processed": "your_file.pdf"
}

Error response (500):

{
  "detail": "Error processing document: <message>"
}

Example API call (PowerShell)

$filePath = "C:\path\to\document.pdf"
$query = "Analyze growth, profitability, risks, and investment implications"

Invoke-RestMethod -Uri "http://127.0.0.1:8000/analyze" `
  -Method Post `
  -Form @{ file = Get-Item $filePath; query = $query }

Deployment Notes

• Backend can be hosted on Render.
• Frontend can be hosted on Netlify.
• Current live frontend: https://financial-document-analyser.netlify.app/
• For production frontend, set:
  • VITE_API_BASE_URL=https://<your-backend>.onrender.com
  • VITE_API_ANALYZE_PATH=/analyze
• Update backend ALLOWED_ORIGINS to include deployed frontend URL.

Security Notes

• Never commit real API keys to git.
• Keep .env files local and use platform environment variables in production.
• Rotate exposed keys immediately if leaked.