This project was developed as part of the AGI Agent Application Hackathon.
It aims to support victims of rental fraud by providing AI-powered legal document analysis and actionable legal guidance by leveraging advanced AI agent technologies.
- ✅ Contract Analysis: Uses Upstage OCR and GPT models to extract and analyze key clauses from rental contracts, identifying hidden risks and unfavorable terms.
- ✅ Actionable Legal Guidance: Users can describe their legal situation, and based on BERT-powered analysis, the system provides relevant laws, precedents, required documents, and step-by-step instructions for legal response.
- ✅ Contextual Q&A: Users can ask questions in natural language, and the system — fine-tuned on legal texts using BERT — returns document-specific, context-aware legal answers.
👉 demo video 👉 live demo website link
Frontend: Streamlit
Backend: Python (custom scripts & API integrations)
Database: FAISS (Vector Store for Legal Document Retrieval)
Others: OpenAI API / Upstage DocAI API/ LangChain / HuggingFace Transformers
📁MINEY/
├── .devcontainer/
├── docs/
├── Predict/
├── readmeiamge/
├── referencnes/
├── 계약서input예시(테스트용)/
├── .gitattributes
├── .gitignore
├── answer.py
├── contract_analysis.py
├── main.py
├── rag_law_current.py
├── README.md
├── requirements.txt
├── risk_assessor.py
├── test.py
├── tutorial1.png
├── tutorial2.png
├── tutorial3.png
└── upstage.py
# Clone the repository
git clone https://github.com/UpstageAI/cookbook/usecase/agi-agent-application/miney.git
# run
streamlit run main.py-
LBox Open Dataset
Korean court rulings and legal clauses used both to fine-tune the BERT model for legal prediction, and to build a FAISS-based retrieval system for RAG (Retrieval-Augmented Generation). -
Custom OCR Samples
Real-world rental contracts, including the official government-standard lease agreement, were collected and anonymized for OCR testing, parsing, and downstream legal analysis.
| Name | Role | GitHub |
|---|---|---|
| Ko Youngkwon | AI Developer | @k0ykwon |
| Seo Suyeon | Frontend Developer | @ellie3413 |
| Yeom Seo Kyung | Backend Developer | @skyyeom |
| Yoon Tae Du | Backend Developer | @taedooit |
| Lim Chaeyoon | Backend Developer | @2022148081 |
Last updated: 2025-04-05
This project is licensed under the MIT license.
See the LICENSE file for more details.
- This project was developed as an MVP during the AGI Agent Application Hackathon.
- All legal responses are generated by AI models and are intended for informational purposes only.
- It is not a substitute for professional legal advice. Please consult a licensed attorney for critical legal decisions.
- We plan to expand the service to other legal domains (e.g., labor law, consumer protection) in the future.