CFA

This project develops a retrieval-augmented multi-agent AI system for drafting Hong Kong IPO prospectuses. Using reverse engineering, it reconstructs disclosure inputs from past prospectuses to build a knowledge base. A multi-agent module simulates drafting, verification, and revision, generating section drafts. Evaluation covers structural completeness, factual grounding, linguistic quality, and professional usability, demonstrating generative AI’s potential in compliance-sensitive domains.