RegRader

Inspiration

For technology companies operating globally, keeping up with ever-evolving IT-related regulations, especially in areas like data privacy, cybersecurity, and AI regulation, is a significant challenge. The legal systems of each country are complex, the languages are different, and the speed of revision is fast, placing a constant burden on legal and compliance departments. This project was born from the idea: "Can we simplify and make predictable the complex IT legal compliance work that crosses borders, using the power of AI?" Instead of "reactively" chasing legal amendments, we aimed to create a forward-looking navigation tool that allows companies to proactively plan by analyzing pre-enforcement regulations and draft-stage information with AI.

What it does

RegRadar is an integrated compliance dashboard for real-time tracking and analysis of IT-related regulations in Japan, the United States, and the European Union. This application provides the following key features:

• Integrated Dashboard: Displays both current laws and future legal amendments in a single, easy-to-use interface.
• Multi-Region Support: Aggregates information from major legal databases in various countries, such as EUR-Lex (EU), e-Gov (Japan), and the Federal Register (US).
• AI-Generated Insights: Utilizes Google's Gemini model to automatically extract "key technical points for the IT department" and "specific impacts on the company" from complex legal documents, providing them in both Japanese and English.
• Future Law Tracking: Monitors laws that have been promulgated but are not yet in effect, as well as bills under deliberation, and visualizes key changes and the number of days until enforcement in an easy-to-understand card format.
• Advanced Filtering and Search: Allows users to quickly access necessary information by filtering by legal field, country/region, impact level, and keywords.

How we built it

RegRadar is built on a modern technology stack.

• Frontend: We used React and TypeScript, with the fast build tool Vite. The UI is styled with Tailwind CSS, and Lucide React is used for icons to create a clean and intuitive interface.
• Backend & Database: We chose Supabase for data storage and backend logic. We designed a detailed schema for the PostgreSQL database to store the master law data, version histories, and AI-generated analysis results.
• Data Sources: Data is acquired from official legal databases of each country (EUR-Lex, e-Gov, Federal Register, etc.) via APIs, XML, and HTML.
• AI Integration: We built a service (aiContentService.ts) that leverages Google's Gemini 1.5 Pro model to generate summaries, key points, corporate impact, and technical requirements from the acquired legal data. This transforms raw legal text into actionable insights.
• Data Migration: A robust data migration system was constructed to store data from external sources into the Supabase database. This ensures data integrity and enables centralized management.

Challenges we ran into

We faced several major hurdles during the development process.

• Diversity of Data Sources: The legal databases of each country had different API specifications, data formats (XML, JSON, HTML, etc.), and update frequencies. The process of absorbing these differences and transforming them into a unified data model (LegalAmendment) was extremely complex.
• Data Reliability and AI Hallucination: Ensuring the reliability of AI-generated information in the legal domain, where accuracy is paramount, was a significant challenge. We went through numerous iterations of prompt engineering to improve the accuracy of the generated information, while also ensuring that links to the original source were always provided so users could verify the primary information.
• Complexity of Asynchronous Operations: Efficiently and stably managing numerous asynchronous processes—such as data acquisition from multiple external APIs, content generation by AI, and database writes—was difficult. While we utilized parallel processing with Promise.all, designing error handling and fallback mechanisms (like displaying mock data on API errors) was particularly time-consuming.
• Database Schema Design: Designing a normalized relational database schema to efficiently manage current laws, future amendments, and their multilingual analysis results was challenging. We particularly struggled with how to relate the "version management" of laws with the "multilingual analysis data".

Accomplishments that we're proud of

We are particularly proud of the following achievements with this project:

• Building the World's First Comprehensive IT Legal Tracking System: We created a system that allows for the cross-sectional monitoring and analysis of IT-related regulations from major economic zones—Japan, the US, and the EU—on a single platform.
• Creating Added Value with AI: By effectively integrating the Gemini API, we were able to go beyond simple collection and display of legal data to generate practical insights that lead to concrete actions on "what the IT department should do".
• Future-Oriented Compliance: We focused on the concept of "future legal amendments" and provided a tool that enables companies to prepare strategically for compliance, rather than being reactive.
• A Robust Data Foundation: Using Supabase and PostgreSQL, we built a scalable and manageable data foundation. The data migration mechanism, in particular, is designed to be flexible enough to accommodate the addition of future data sources.

What we learned

Through this project, we learned a great deal.

• The Importance of the Data Model: We realized that when dealing with heterogeneous external data, designing a flexible and normalized internal data model at the outset is crucial for the success of the entire project.
• AI is Not a "Magic Wand": AI is a powerful tool, but it is not a panacea. We learned that in highly specialized domains, a hybrid approach that combines AI-generated results with expert review and verification against primary sources is essential.
• The Value of UI/UX: Especially for applications that handle complex information, we reaffirmed the importance of an intuitive and visual dashboard (e.g., statistical info cards) that allows users to grasp the situation at a glance.
• Separation of Development and Production Environments: Incorporating mechanisms for switching API keys and data sources in the early stages (e.g., switching between real and mock data in the useLegalData hook) made the later development process smoother.

What's next for RegRadar

The future of RegRadar is full of possibilities. As next steps, we are planning the following enhancements:

• Expansion of Target Countries/Regions: In addition to the current Japan, US, and EU, we will expand coverage to include the UK, Canada, Australia, and key countries in the Asia-Pacific region (such as Singapore and South Korea).
• Advancement of AI Analysis: We are considering fine-tuning AI models specialized in specific legal areas to enable more expert legal interpretation and industry-specific impact analysis. We also want to add a feature to compare and analyze multiple draft amendments.
• Enhanced Personalization and Notification Functions: Users will be able to register their areas of legal interest and regions, and receive real-time notifications via email or in-app when there are developments in related legal amendments.
• Compliance Project Management: We aim to integrate simple project management functions within the application, allowing for task management and progress tracking to meet each legal requirement.
• Reporting Functionality: We will add a feature to automatically generate reports summarizing the status of tracked legislation and a company's own compliance status, exportable as PDF or CSV.

Built With

• e-gov
• eur-lex
• google-gemini
• javascript
• postgresql
• react
• sql
• supabase
• typescript
• vite