⚖️ AI Contract Risk Analyzer

AI-Powered Legal Document Intelligence for Faster, Smarter Contract Review

Upload a contract. Get clause-level risk analysis in minutes, not hours.

Features • Workflow • Architecture • Installation • Skills Demonstrated

---

📖 Introduction

AI Contract Risk Analyzer is an AI-powered legal document analysis platform that automates the first-pass review of commercial contracts. It extracts and classifies key legal clauses, flags missing protections, scores overall contract risk, and generates AI-driven legal analysis — all through a clean, interactive Streamlit interface, powered by a locally hosted Llama 3.2 model via Ollama.

Built by a practicing corporate lawyer, the tool reflects how contract review actually happens in practice — not a generic document-summarization demo, but a workflow modeled on real due diligence and transactional review.

🏢 In production use: Formally adopted by Ch. Jehangir Law Firm to support real client contract review work.

---

🧩 Problem Statement

Manual contract review is one of the most time-intensive, expensive parts of legal and transactional work:

• A single commercial contract can take a lawyer 30–90 minutes to review manually for key clauses and risk exposure
• Junior lawyers and paralegals spend significant billable time on repetitive, first-pass review rather than higher-value analysis
• Manual review is prone to inconsistency — different reviewers catch different issues, and missing clauses (e.g., no indemnity provision, no governing law clause) are easy to overlook under time pressure
• Clients and deal teams need fast turnaround, especially during active M&A due diligence, when dozens of contracts must be reviewed against a tight timeline

This creates a clear bottleneck: legal review that should take minutes often takes hours, and cost scales linearly with document volume.

💡 Solution

AI Contract Risk Analyzer addresses this bottleneck by automating the first-pass review layer:

• Extracts text from PDF and DOCX contracts automatically
• Detects the presence and content of key legal clauses
• Identifies gaps — clauses a well-drafted contract should have but doesn't
• Scores risk using a rules-based and AI-assisted engine
• Generates legal reasoning using a locally hosted LLM (Llama 3.2), so no contract data ever leaves the local environment
• Produces a report a lawyer can review, verify, and act on in minutes

The result: lawyers spend their time on judgment and negotiation, not on manually scanning for clauses that a first pass can catch automatically.

---

✨ Features

Category	Capability
📄 Document Ingestion	Upload contracts in PDF or DOCX format
🔍 Text Extraction	Automatic, structure-aware text extraction
🏷️ Clause Detection	Identifies and classifies key legal clauses
⚠️ Missing Clause Detection	Flags standard clauses absent from the contract
📊 Contract Classification	Categorizes contract type based on content
🎯 Risk Scoring Engine	Quantifies overall contract risk exposure
🤖 AI Legal Reasoning	Locally hosted Llama 3.2 (via Ollama) generates contextual legal analysis
📈 Executive Dashboard	KPI cards summarizing key contract metrics
📉 Risk Visualization	Interactive charts of clause and risk statistics
📥 Downloadable Reports	Exportable, shareable risk analysis reports
🖥️ Clean Interface	Streamlit-based UI, no legal or technical background required to operate

🔑 Legal Clauses Detected

✅ Confidentiality	✅ Termination	✅ Indemnity
✅ Liability	✅ Force Majeure	✅ Governing Law
✅ Arbitration	✅ Payment	✅ Intellectual Property
✅ Non-Compete

---

🔄 Workflow

flowchart TD
    A[📄 Upload Contract<br/>PDF / DOCX] --> B[🔤 Extract Text]
    B --> C[🏷️ Detect Important Clauses]
    C --> D[⚠️ Identify Missing Clauses]
    D --> E[🎯 Calculate Risk Score]
    E --> F[🤖 AI Legal Analysis<br/>Llama 3.2 via Ollama]
    F --> G[📋 Generate Professional Report]
    G --> H[📥 Download & Review]

    style A fill:#3776AB,color:#fff
    style F fill:#0467DF,color:#fff
    style G fill:#FF4B4B,color:#fff
    style H fill:#2E7D32,color:#fff

Loading

---

📸 Screenshots

Screenshots below illustrate the core application views.

Dashboard	Upload	Report

---

🏗️ System Architecture

flowchart LR
    subgraph Client["🖥️ Streamlit Interface"]
        UI[User Uploads Contract]
    end

    subgraph Core["⚙️ Processing Core"]
        TE[text_extractor.py]
        CD[clause_detector.py]
        RE[risk_engine.py]
        AA[ai_analyzer.py]
    end

    subgraph AI["🤖 AI Layer"]
        OL[Ollama Runtime]
        LM[Llama 3.2 Model]
    end

    subgraph Output["📊 Output Layer"]
        DASH[Executive Dashboard]
        REP[Downloadable Report]
    end

    UI --> TE --> CD --> RE
    RE --> AA
    AA <--> OL
    OL <--> LM
    AA --> DASH
    AA --> REP

    style Client fill:#FF4B4B,color:#fff
    style AI fill:#0467DF,color:#fff
    style Output fill:#2E7D32,color:#fff

Loading

Module Responsibilities

Module	Responsibility
text_extractor.py	Parses PDF/DOCX files and extracts clean, structured text for downstream processing
clause_detector.py	Scans extracted text to identify, classify, and extract the content of key legal clauses
risk_engine.py	Applies scoring logic across detected clauses, missing clauses, and contract characteristics to produce a risk score
ai_analyzer.py	Sends structured clause and risk data to the locally hosted Llama 3.2 model via Ollama and returns contextual legal analysis and an executive summary

---

🛠️ Technology Stack

Layer	Technology
Frontend	Streamlit
Backend	Python
AI Runtime	Ollama
AI Model	Llama 3.2
Data Handling	Pandas
Visualization	Plotly
PDF Parsing	pdfplumber
DOCX Parsing	python-docx

---

⚙️ Installation

Prerequisites

• Python 3.10 or higher
• Ollama installed locally
• Git

Step 1 — Clone the Repository

git clone https://github.com/<your-username>/ai-contract-risk-analyzer.git
cd ai-contract-risk-analyzer

Step 2 — Create a Virtual Environment

python -m venv venv
source venv/bin/activate      # On Windows: venv\Scripts\activate

Step 3 — Install Dependencies

pip install -r requirements.txt

---

🚀 Running Locally

Step 1 — Install and Start Ollama

# Install Ollama (if not already installed)
# https://ollama.com/download

# Pull the Llama 3.2 model
ollama pull llama3.2

# Ollama runs automatically as a local service after installation

Step 2 — Launch the Application

streamlit run app.py

Step 3 — Open in Browser

Navigate to http://localhost:8501 and upload a contract to begin analysis.

---

📁 Folder Structure

ai-contract-risk-analyzer/
│
├── app.py                     # Main Streamlit application entry point
├── requirements.txt           # Python dependencies
├── README.md                  # Project documentation
│
├── utils/
│   ├── text_extractor.py      # PDF/DOCX text extraction
│   ├── clause_detector.py     # Clause detection and classification
│   ├── risk_engine.py         # Risk scoring logic
│   └── ai_analyzer.py         # Llama 3.2 / Ollama integration
│
├── screenshots/
│   ├── dashboard.png
│   ├── upload.png
│   └── report.png
│
└── LICENSE

---

🔮 Future Improvements

• 🌐 Support for multi-language contracts
• 🖨️ OCR support for scanned/image-based contracts
• 📚 Multi-document simultaneous review
• 🔀 Clause-to-clause comparison across contract versions
• ✍️ AI-assisted redlining and clause suggestions
• ☁️ Cloud deployment with secure multi-user access

---

🎓 Skills Demonstrated

⚖️ Legal Skills Contract review methodology Clause identification & risk assessment Due diligence workflows Corporate & commercial law application

🤖 AI Skills Local LLM deployment (Ollama) Prompt design for legal reasoning AI-assisted document analysis Applied NLP for legal text

💻 Programming Skills Python application development Modular software architecture Streamlit UI development PDF/DOCX parsing pipelines

📈 Business Skills Legal workflow automation Process efficiency analysis Product thinking for legal tech Stakeholder-focused design

---

🧭 Key Learning Outcomes

Building AI Contract Risk Analyzer required bridging legal domain expertise with applied software engineering — translating an informal, judgment-based legal review process into a structured, repeatable pipeline. Key outcomes included learning to design modular Python architecture that separates extraction, detection, scoring, and AI reasoning into independently maintainable components; integrating a locally hosted LLM into a production workflow, balancing analysis quality against latency and reliability; and developing product judgment around what a legal end user actually needs from an AI tool's output — clear, actionable, and verifiable, rather than a black-box result the user has to blindly trust.

---

⚠️ Disclaimer

This project is developed for educational and research purposes only. It is a portfolio and applied-learning project demonstrating the intersection of legal practice and artificial intelligence. It does not constitute legal advice, and its output should not be relied upon as a substitute for review by a qualified legal professional. Any use of this tool in a professional legal context should be treated as a supplementary review aid, subject to verification by a licensed lawyer.

---

📄 License

This project is licensed under the MIT License — see the LICENSE file for details.

---

Built by Ali Asad Ullah Junior Associate, Ch. Jehangir Law Firm | Corporate Law × AI × Finance