AI Experiments
Make Prototype Test #1
For this experiment, I wanted to animate a simple date picker and test Make for how much context it would need for a simple animation.
GOAL
Test how much context Make would need for a simple prototype.
MY ROLE
Figma Designer
AI USED IN TESTING
Gemini for results evaluation
Figma Make for prototype animation
Method A
I wanted to see what it would do with a nested Figma Component and minimal context in the initial prompt.
It created a simple form as requested with functional inputs. However it did not respect the nested component props so the visuals (it used the wrong prop) and animation was not as expected nor did it make the inputs and component fit together.
The results were less than expected.
Method B
After being underwhelmed by the results of Method A, I used Gemini to evaluate the prompt strategy, explain why Make ignored the nested prompts and to suggest a better prompt structure.
I created some simple mock ups of the form in Figma using the date picker component. I put these and the examples of the nested component states for the dates in a frame. I used the frame as the starting point for Make and used a significantly more explicit and detailed prompt.
The results were much better this time. It produced a working form that visually matched the design and the animation was visually as expected. Specifically the inputs fit with the date component and the correct props were used.
So it appears Make requires explicit instructions and full mockups for animation.
Make Prototype Test #2
This experiment was a bit more ambitious. I wanted to test how Make would perform with a proof of concept prototype that was not designed.
GOAL
Test Make‘s ability to create a Proof of Concept
MY ROLE
Product Manager
QA (functionality)
Design Reviewer
GEMINI’S ROLE
Senior Engineer for AI architecture
MAKE’s ROLE
Junior Engineer for execution