An AI-powered regulatory copilot designed to assist entrepreneurs in navigating company formation, licensing, visa procedures, and compliance workflows in Meydan Free Zone, Dubai.
This project demonstrates how Retrieval-Augmented Generation (RAG) can be applied to build domain-specific AI systems that provide grounded, context-aware guidance for real-world regulatory processes.
Entrepreneurs setting up companies in UAE free zones often face:
- Fragmented regulatory information across multiple sources
- Confusion around visa eligibility and sequencing
- Delays due to incomplete documentation
- Lack of clear compliance workflow visibility
- Dependence on consultants for basic procedural guidance
Even though free zones provide digital setup portals, founders still struggle to understand the end-to-end regulatory journey.
The Meydan Free Zone AI Compliance Copilot provides:
- Context-aware regulatory guidance using curated knowledge sources
- Step-by-step support for company formation workflows
- AI-assisted understanding of licensing and visa procedures
- Semantic retrieval of relevant compliance information
- Grounded responses instead of generic chatbot outputs
This transforms static regulatory documentation into an interactive decision-support system.
The system follows a Retrieval-Augmented Generation (RAG) architecture.
- Regulatory documents are collected and stored as a domain knowledge base
- Documents are split into semantic chunks for efficient retrieval
- Each chunk is converted into vector embeddings
- A FAISS vector index enables fast similarity search
- User queries are embedded and matched against the vector store
- Relevant context is retrieved
- The language model generates a grounded, contextual response
Traditional chatbots generate generic responses.
RAG enables:
- Domain grounding
- Regulatory accuracy
- Reduced hallucinations
- Explainable AI behaviour
- Scalable knowledge updates
This approach is suitable for enterprise regulatory automation systems.
Below is the Streamlit interface demonstrating the AI Compliance Copilot.
User Question
How do I register a company in Meydan Free Zone?
AI Behaviour
- Retrieves licensing workflow context
- Identifies required regulatory steps
- Generates structured procedural guidance
- Python
- Streamlit
- OpenAI API
- FAISS Vector Database
- NumPy
- Retrieval-Augmented Generation (RAG)
- Semantic Search
- Vector Embeddings
- Context Grounding
meydan-ai-compliance-copilot
│
├── app
│ └── app.py
│
├── docs
│ ├── faq.txt
│ ├── license_renewal.txt
│ ├── setup_process.txt
│ └── visa_challenges.txt
│
├── screenshots
│ ├── ui.png
│ └── example-answer.png
│
├── architecture.png
├── requirements.txt
└── README.md
Clone the repository:
git clone https://github.com/yourusername/meydan-ai-compliance-copilot.git
Navigate into the project directory:
cd meydan-ai-compliance-copilot
Install the required dependencies:
pip install -r requirements.txt
Start the Streamlit application:
python -m streamlit run app/app.py
The application will open in your browser at:
http://localhost:8501
Enter your OpenAI API key and start asking questions.
- How do I register a company in Meydan Free Zone?
- What documents are required for trade license approval?
- How does visa allocation work for founders?
- What compliance steps must startups follow?
- How does license renewal work in UAE free zones?
- Integration with official free-zone APIs
- Multilingual support (Arabic + English)
- Workflow automation instead of static guidance
- Live regulatory data updates
- Support for multiple UAE free zones
- Deployment as enterprise SaaS compliance platform
- Entrepreneurs planning UAE company formation
- Free-zone authorities building AI support systems
- Business consultants automating regulatory guidance
- AI researchers exploring domain-specific RAG systems
Ashish Seru
MSc Artificial Intelligence
De Montfort University Dubai
This project is intended for educational, research, and demonstration purposes.
This project is intended for educational, research, and demonstration purposes.