Generative AI Integration in a Content Management System : Aligning Persuasive Design with User Needs in Automated Document Summarisation

This thesis investigates the integration of a generative AI (GenAI) model into a Content Management System (CMS) to fulfil user needs in alignment with persuasive design methods. Conducted in collaboration with Swace Digital and a large non-governmental organisation (NGO), the project centres on AI-assisted abstract generation for uploaded PDF documents. Guided by Fogg’s Behavior Model and Self-Determination Theory, a persuasive design approach was adopted. User needs were identified through semi-structured interviews and analysed using thematic analysis, revealing autonomy (e.g user control) and simplicity to be important qualities for an integrated GenAI tool. The solution was implemented as a WordPress plugin that communicates with an external serverless API hosted on AWS Lambda, which acts as a middleware to interface with OpenAI’s GPT-4o-mini model. It was evaluated through think-aloud usability tests and interviews. The results indicate that collaboration with GenAI in summarisation can improve workflow efficiency and perceived value without compromising user autonomy—though increased autonomy also heightens the demand for simplicity. This work contributes to the fields of persuasive design and human-AI collaboration by showing that successful adoption depends on balancing efficiency gains with user needs.