Hessel Frank Mullemeister
Compliance without compromise: a private, high-precision AI chatbot with RAG for answering questions about Dutch financial regulation.
Rel. Gianvito Urgese. Politecnico di Torino, Corso di laurea magistrale in Digital Skills For Sustainable Societal Transitions, 2026
|
Preview |
PDF (Tesi_di_laurea)
- Tesi
Licenza: Creative Commons Attribution Non-commercial No Derivatives. Download (983kB) | Preview |
|
|
Archive (ZIP) (Documenti_allegati)
- Altro
Licenza: Creative Commons Attribution Non-commercial No Derivatives. Download (26MB) |
Abstract
A Dutch trust office offering compliance as a service to companies under dense, layered anti-money-laundering and integrity obligations (Wwft, Wtt 2018, Sanctiewet 1977) needs deep, specialised knowledge of legislation, supervisory guidance and case law. Identifying compliance gaps means searching through statutes, implementing decrees, guidance and case law: slow, expert work in which a mistake can carry regulatory and reputational liability. Popular large language models offer faster research, but hallucination and limited auditability are key drawbacks. Strict confidentiality standards also prevent uploading sensitive client ownership structures and source-of-funds material to a public cloud model. A market scan confirms that no existing product combines Dutch trust-office coverage, private deployment and client-document handling in one tool.
This thesis asks whether it is possible to build a private, bilingual compliance assistant, designed to perform best in Dutch, that is useful enough to help and transparent enough to be auditable
Relatori
Anno Accademico
Tipo di pubblicazione
Numero di pagine
Corso di laurea
Classe di laurea
Aziende collaboratrici
URI
![]() |
Modifica (riservato agli operatori) |
