AI-Powered Financial Wellness

First Bank | Amsterdam

Goal

To design a financial wellness experience that increases engagement and supports smarter financial decisions using AI, measured by a 20% increase in interaction with insights, improved task success rates, and higher SUS usability scores.

My role

I led the research, product discovery, design prototyping, and testing for this concept, using Gemini Deep Research and Lovable AI to inform the UI experience. I collaborated cross-functionally with Product and Design to align on vision and execution.

Impact

Insights from this concept are now shaping client conversations around a complete and more personalized Financial Wellness overhaul.

Before and After

Where we started

During the Retail Banking Revamp, most of our focus was on improving the app’s dashboard and overall information architecture. We streamlined navigation and reduced friction across core user flows.

At the time, the Financial Insights page was largely out of scope. It existed as a generic screen with limited insights or personalization — and user engagement was low. We saw an opportunity to revisit this underutilized area and reimagine how insights could feel more actionable, emotionally supportive, and easy to act on.

Our first steps

  • Revisited existing analytics to understand how users were interacting with the Financial Insights page.

  • Reviewed prior research from the Retail Banking Revamp, focusing on navigation, dashboard usage, and content hierarchy.

  • Began experimenting with our internal AI toolkit: Notebook LM, Gemini Deep Research, Lovable AI, and Figma Make.

Discovery

Here’s what I did:

  • Survey review: I revisited a previously run financial wellness survey with 397 participants across the US, UK, and Canada to identify what people expect from financial dashboards. Key findings shaped early decisions around clarity, forecasting, and actionable guidance.

  • Segment & persona alignment: I used Gemini Deep Research to surface behavioral segments and mapped these against existing personas to clarify needs, goals, and pain points across different user types.

  • Prompt design & ideation: I created a structured prompt combining survey themes, behavioral segments, and tone considerations, and input it into Lovable AI to generate actionable nudges.

  • Rapid concept iteration: I translated the AI nudges into a first concept prototype, integrating forecasting, debt tracking, and contextual guidance. Internal feedback and helped refine this direction.

I explored multiple ways to organize and prioritize financial insights. Each layout tested different strategies for reducing overwhelm and guiding user focus.

Concept exploration: layout variations

Lovable AI prompt for generating insight card components

We moved forward with a combination of Concept 1 and Concept 2, as it offered clear direction for users while maintaining focused access to specific topics.

Concept design

Design considerations

Designing AI-powered financial insights raised critical questions:

  • What if there’s no data or the wrong data?

  • How do we build trust in what the user is seeing?

  • Can users understand and act on the insight?

  • Are we surfacing the right amount of information?

  • Could this insight cause anxiety or harm?

  • Is it clear what the next step is?

These questions shaped our copy, fallback states, and overall approach to AI transparency and emotional tone.

Impact

Research on the redesigned Financial Wellness dashboard has so far indicated:

  • Strong perceived value and emotional resonance. Participants described it as “very helpful” for managing their finances, highlighting its clarity and depth compared to other apps.

• Strategic influence. Insights from this concept are now informing client conversations around a full-scale Financial Wellness overhaul.

• Team clarity on AI integration. The project helped clarify what an end-to-end AI design process can look like and how to integrate AI in a way that augments, rather than disrupts or overwhelms, the user journey.