ComplianceCopilot

AI-Driven Compliance Analyzer for Financial Documents

Regulation-Grounded • LLM-Powered • Audit-Ready

---

---

Overview

ComplianceCopilot is an AI-powered system that analyzes financial documents such as:

• Loan Agreements
• Terms & Conditions
• Key Fact Statements (KFS)

It compares them against RBI/SEBI guidelines using:

• Document parsing
• Clause classification
• Regulation retrieval
• LLM-based reasoning
• Risk scoring
• Explainable reporting

The goal is to automate and streamline compliance reviews with accuracy, speed, and clarity.

---

Features

📝 Document Ingestion

• Upload PDF, DOCX, or TXT files
• Extracts clean text
• Splits into meaningful clauses

🔎 Clause Classification

Automatically identifies:

• Pricing / APR
• Fees & Charges
• Foreclosure
• Grievance Redressal
• KYC / Documentation
• Security
• Default & Recovery

📚 Regulation Retrieval

• Fetches the most relevant regulatory snippets
• Uses TF-IDF or embeddings (expandable)

🤖 LLM-Powered Gap Detection

• Compares document clauses with regulatory expectations
• Generates:
  • Compliance status
  • Issues
  • Severity
  • Suggested remediation
  • Citations
• Automatic fallback to heuristic comparator if needed

🔥 Risk Scoring

• Severity-based risk calculation
• Document-level overall score

🖥 Streamlit UI

• File upload
• “Process File” button
• Progress display
• Comparator mode switch
• Detailed issue breakdown
• Report export

🧪 Test Suite

• Comparator tests using pytest

---

Architecture

📂 ComplianceCopilot
│
├── ingestion/            # PDF/DOCX loaders, text cleaning
├── clause_analyzer/      # Clause splitting & classification
├── rag_engine/
│     ├── retriever.py    # Retrieval logic
│     ├── comparator.py   # LLM + heuristic comparator
│     └── prompts.py      # Prompt templates
├── risk_engine/          # Risk scoring
├── report/               # Report generator
├── tests/                # Pytest suite
└── app.py                # Streamlit frontend

---

Installation

1️⃣ Create virtual environment

python3 -m venv venv
source venv/bin/activate

2️⃣ Install dependencies

pip install -r requirements.txt

3️⃣ Optional: enable LLM mode

export OPENAI_API_KEY="your-key-here"

4️⃣ Run the application

streamlit run app.py

---

Running Tests

pytest -q

---

Example Issue Output

[HIGH] APR not clearly disclosed.
Regulation: RBI Fair Practices Code
Reason: Customers must be informed of total cost of credit.
Fix: Include APR + example calculation.

---

Roadmap

• OCR for scanned PDFs
• SEBI + IRDAI regulation expansion
• Embedding-based retriever
• Version comparison for documents
• Export full PDF report
• Deploy to cloud (Render/Railway)

---

Contributing

Pull requests are welcome. Open an issue before major changes.

---

License

Distributed under the MIT License.

---

Built with passion to simplify compliance workflows.