The Challenge

While working on a banking system, I needed to design a loan restructuring form that functioned as a dynamic calculator. Creating a functional prototype for user testing was complex and time-consuming using traditional tools.

 

​

Step 1: Writing the prompt with ChatGPT

I started by asking ChatGPT to create a Figma Make prompt:

• Explained the form's logic and field constraints

• Uploaded a screenshot of the initial design

• Defined the prototype's purpose and testing goals

Key insight: I invested significant time in the initial prompt since everything builds on it.

 

​

Step 2: Implementation in Figma Make

I copied the prompt to Figma Make and uploaded the form as a base frame. The result was an almost perfectly functioning form. I then tested edge cases and fixed several logic issues.

 

​

Step 3: Interview Preparation

In the same ChatGPT conversation, I requested a structured interview flow focusing on:

• Input field relationships with calculator visibility

• Scanning and readability of specific areas

• User behavior patterns

This gave me confidence I wasn't missing important aspects.

 

​

The User Testing

Our UX researcher recruited a Canadian banker for remote testing. Due to security constraints, I couldn't share the prototype directly, I showed my screen while the user guided me where to click.

Critical principle: I smiled and nodded whenever the user gave negative feedback to make them feel this was exactly the feedback we wanted.

The session was automatically transcribed using Google Gemini.

 

​

Surprising Insights

The user revealed issues I would never have discovered:

1. Completely ignored an important button in the form

2. Misunderstood a component I thought was obvious, the loan header contained the total amount, but because of a nearby "overdue" chip, the user thought it showed the overdue amount

Key realization: I was trapped in the system's conventions and couldn't see it through fresh eyes.

 

​

Processing Results

I immediately documented all feedback on the Figma file, then asked ChatGPT to summarize the Gemini transcript. This created an excellent summary divided into:

1. Facts - What the user did and said

2. Insights - What I understood and could improve

 

​

Impact

• Prototype creation: Reduced from 2 weeks to 30 minutes

• Interview prep and analysis: Significantly faster

• Focus shift: More time for the exciting parts, observing live user feedback and creative problem-solving

 

​

Conclusion

AI tools in user testing don't replace design thinking. They free us up to focus on what truly matters: understanding users and creative problem-solving. The time saved on technical and administrative tasks allows designers to invest more in the human insights that drive great products.