Building virtual pension participants with AI

Pensions are often such a complex and difficult subject that, as a participant, you easily put it off until tomorrow. Project Mañana is about improving user journeys within the pension world. By experimenting with new technology, storytelling, and design, we can make pensions more enjoyable, understandable, and, above all, more relevant for participants. During the first hackathon, our designers and developers immediately tackled a cool research question!

First, we built our own workflows using tools like Activepieces and n8n. We then linked these to a custom-built AI assistant. By feeding the assistant with data from various sources, it can suggest different visitors to your pension platform.

Based on the virtual participant's characteristics, such as reading level, financial knowledge, or digital skills, you can determine how comprehensible the content will be for them. The AI assistant analyzes whether the website's content aligns with the participant's understanding. All content can be customized for each individual, making it more understandable for everyone.

Besides using a virtual assistant to create personas that can assess and improve UX and content, there are many other conceivable applications.

Consider, for example, a brand police assistant that scans your content to ensure it complies with your organization's branding regulations. Or an assistant as a compliance checker, assessing your compliance with laws and regulations such as Accessibility, GDPR, or DORA.

AI always generates output, but often of insufficient quality: a six. The biggest challenge remains to raise it from this six to a consistent nine. That requires more than prompt engineering: by breaking down tasks, for example, you can use much more specific models and validate much more targeted output. A great starting point for further experiments!