Decoding the DOA – a case study in simplifying governance

Our team were engaged to identify the most effective technology solution through a POC. We initially explored Large Language Models (LLMs) to test whether Generative AI, grounded in the governance documentation, could deliver reliable results. However, the complexity and formatting of the source documents created significant interpretation challenges for LLMs without extensive reworking or prompt engineering.

The breakthrough: business rules methodology

To deliver the accuracy and consistency the organisation needed, we developed a structured methodology that translated governance documentation into nearly 1,900 discrete "Business Rules.”

Using "IF/THEN" logic, we converted policy requirements into clear, actionable statements captured in Excel—ready for codification in a Business Rules Engine (BRE).

We designed a modular output structure with reusable components, enabling rapid drafting, high flexibility, and straightforward maintenance. This was critical: governance documents change frequently, and legal team members responsible for updates aren't always technical. Our approach allows targeted updates without rewriting entire rule sets.

We built a prototype chatbot assistant powered by a BRE, with 252 codified Business Rules covering one approval process. The POC successfully demonstrated that our methodology could provide business users with accurate, tailored guidance and direct access to relevant documentation.

Following the successful POC, we were engaged for a three-month technology enablement project. Our team reviewed and analysed detailed governance materials, transforming them into user journey maps, comprehensive Business Rule sets, and modular output components.