We built CompliAI for professionals, students, and anyone who struggles to understand complex legal and compliance documents. Organizations risk costly violations due to unclear policies, while students and non-experts often find legal language difficult to interpret. Our goal was to bridge this gap by making compliance analysis and legal understanding accessible, actionable, and fast.
We used AI (Claude) as the core engine to analyze documents, identify compliance risks, highlight problematic clauses, and suggest improvements. In Risk Mode, Claude generates structured outputs like risk scores, severity levels, relevant regulations, and recommendations. In Student Mode, it simplifies the same content into summaries, key points, and plain-English explanations. We also leveraged AI to rewrite policies, enabling a before-and-after comparison that demonstrates reduced risk.
One challenge is that AI outputs can sometimes be inconsistent or not strictly structured, especially when expecting JSON responses. To address this, we enforced strict prompting, added validation and error handling on the backend, and provided fallback debugging outputs. Another risk is over-reliance on AI for legal interpretation, so we designed the system to present suggestions and insights rather than definitive legal advice.
With more time, we would enhance the platform by adding risk impact simulation (e.g., estimating potential fines or business impact), interactive explanations for highlighted clauses, and support for multiple regulatory frameworks. We would also improve UI/UX with visual dashboards and expand document support for real-world enterprise use.