A product design exploration investigating how AI can accelerate ideation and interface production while keeping strategic product decisions in the hands of the designer.
This was an AI-assisted product design exploration created with Codex.
I used Codex to explore interface directions, extend screens, test responsive layouts and generate component variations. But the project was not a clean, one-click AI workflow.
The process was messy. Codex helped with speed, but it also introduced inconsistencies, broke layouts and required constant design review. Because of that, I decided to frame this case study honestly: not as a perfect banking app, but as an exploration of a product idea developed through human decision-making and AI-assisted iteration.
A banking app should not only show what happened. It can also help users understand what is likely to happen next.
This project started with a very simple observation:
A bank balance does not tell the full story.
Seeing £2,400 in an account does not automatically mean that £2,400 is safe to spend. There may be upcoming bills, subscriptions, scheduled payments, rent, credit card repayments or personal saving goals waiting in the background.
I wanted to explore how a banking dashboard could move beyond showing static account data and instead help users understand their financial situation more clearly.
The core question became:
How much can I safely spend, and how much could I realistically save?
PROBLEM
USER GOALS
ASSUMPTIONS
PAIN POINTS
BUSINESS GOAL
AI-assisted workflow
Codex was used to support the design process, especially during exploration.
It helped generate:
layout variations,
responsive versions,
component directions,
additional states,
early product flows.
However, AI was not able to own the design process.
It often needed strong constraints. When the scope was too broad, the output became inconsistent. It sometimes changed components outside the requested area, introduced layout issues or expanded the product in directions that were not central to the concept.
Because of that, my role was to constantly review, reduce and redirect the work.
The most important design decisions were not generated by AI. They came from evaluating what supported the Smart Savings concept and what distracted from it.
What I learned
This project taught me two things at the same time.
First, the product idea became clearer when I stopped trying to design a full banking app and focused on one valuable user problem.
Second, AI can support fast exploration, but it needs strong direction.
Codex was useful for generating options and speeding up production work, but it did not replace product thinking. It required review, constraints and sometimes a decision to stop expanding.
The biggest lesson was:
AI can help create possibilities, but the designer is still responsible for choosing the right direction.
Reflection
This project is not presented as a flawless banking product.
It is an exploration of a specific product idea: helping users understand what they can safely spend and realistically save.
The process was imperfect, especially when working with AI-generated layouts and components. But that imperfection became part of the learning.
The final value of the project is the Smart Savings concept and the design thinking behind it: moving from passive banking data toward proactive financial clarity.
