Experiment 02

AI as a Speed Layer for Design Thinking

I took the initiative to explore Figma Make on a company project because the team needed to move quickly. The goal was to build a functional prototype, test different models, and use efficient prompts to create something strong enough for PM review and customer feedback before the product was built.

Fast timeline

Quick exploration and faster concept movement.

Design ownership

AI helped with speed, but the decisions stayed with me.

Model exploration

I compared how different models interpreted the same concept.

Customer validation

The prototype made early interaction feedback easier to get.

Why I explored it

Faster concept work, without losing design judgment.

The project moved fast, so I needed a way to stay in sync with the product manager while exploring multiple concepts. Figma Make gave me that speed. It let me branch into different prototype directions, compare them early, and reduce the time it would normally take to iterate many screens manually.

That speed mattered, but it did not replace the design work. The product decisions, interaction logic, and final direction still came from me. The prototype only became useful when the prompts were clear enough and the direction was strong enough to create something that PMs and customers could actually react to.

How it worked

Prompting, model choice, and branching ideas

01

Start with a fast-moving project

The project needed quick movement between me and the product manager, so I explored Figma Make to keep ideas moving without waiting on fully manual screen iteration.

02

Build a functional prototype

The goal was not to make something decorative. I wanted a prototype that was functional enough to talk through the interaction and show how the product would behave.

03

Compare models and directions

I tested different models and branched into separate directions to see which version handled the concept best and produced the clearest interaction story.

04

Use it for A/B testing

With real customers, the prototype helped compare options and gather feedback before the product existed, which made the interaction easier to understand and evaluate.

Model comparison

Different models handled the same prompt differently

Different models

I tried multiple models to compare how each one handled interaction depth, speed, and the quality of the generated prototype.

Efficient prompts

I learned that good output depended on clear direction, specific prompts, and enough context for the prototype to be presentable to PMs and customers.

Branching prototypes

Instead of forcing one direction too early, I branched into separate prototype ideas and compared them while the project was still moving quickly.

What I learned

AI is changing how design moves forward

AI gave speed

Figma Make helped reduce the time it would normally take to iterate across many screens and interaction states.

Design decisions stayed with me

The product thinking, interaction choices, and final direction still came from my judgment as the designer.

Customers understood the interaction earlier

The prototype made it easier to explain the flow before the product was built, which led to better feedback and stronger decisions.

A shift for other designers

This was also a useful pivot for other designers to explore Figma Make as part of early concept testing.

Before build

Better feedback came from showing the interaction early.

The prototype helped people understand the interaction before the actual product was built. That created a big shift in how design could be used in its early state: instead of waiting for a finished product, I could test the concept, collect real customer feedback, and make better decisions sooner.

Closing thought

AI helped me move faster, but the design stayed mine.

This was also a useful pivot for other designers. It showed that Figma Make is not just a shortcut for generating screens. It can be a practical way to test ideas, compare concepts, and bring a more interactive version of the thinking into the room earlier.

The main lesson was simple: AI can speed up design exploration, but it still needs clear prompting, strong direction, and real product judgment to become something worth presenting to PMs and customers.