Testing Figma Make

UX Research

UX Design

Interaction Design

Prototyping

Design Operations

Overview

Figma Make is an AI-driven design and prototyping tool intended to accelerate ideation and reduce time to interactive output. With its release, design leadership at BNY asked the team to assess whether it could responsibly support UX work in a regulated financial environment.

I led an evaluation of Figma Make, testing its ability to generate early UX concepts from prompts and translate existing designs into high-fidelity, interactive prototypes. The goal was to determine where the tool added real value, where it introduced risk, and whether it warranted broader adoption.

My Role

As the UX lead for this work, I owned the evaluation end to end. I defined the scope, ran hands-on experiments across representative product flows, and synthesized findings into clear guidance for the design team.

Responsibilities included:

  • Evaluation strategy and test scoping

  • Hands-on prototyping and experimentation

  • Analysis of performance, reliability, and limitations

  • Recommendations for responsible AI-assisted design

Overview

Figma Make is an AI-driven design and prototyping tool intended to accelerate ideation and reduce time to interactive output. With its release, design leadership at BNY asked the team to assess whether it could responsibly support UX work in a regulated financial environment.

I led an evaluation of Figma Make, testing its ability to generate early UX concepts from prompts and translate existing designs into high-fidelity, interactive prototypes. The goal was to determine where the tool added real value, where it introduced risk, and whether it warranted broader adoption.

My Role

As the UX lead for this work, I owned the evaluation end to end. I defined the scope, ran hands-on experiments across representative product flows, and synthesized findings into clear guidance for the design team.

Responsibilities included:

  • Evaluation strategy and test scoping

  • Hands-on prototyping and experimentation

  • Analysis of performance, reliability, and limitations

  • Recommendations for responsible AI-assisted design

Overview

Figma Make is an AI-driven design and prototyping tool intended to accelerate ideation and reduce time to interactive output. With its release, design leadership at BNY asked the team to assess whether it could responsibly support UX work in a regulated financial environment.

I led an evaluation of Figma Make, testing its ability to generate early UX concepts from prompts and translate existing designs into high-fidelity, interactive prototypes. The goal was to determine where the tool added real value, where it introduced risk, and whether it warranted broader adoption.

My Role

As the UX lead for this work, I owned the evaluation end to end. I defined the scope, ran hands-on experiments across representative product flows, and synthesized findings into clear guidance for the design team.

Responsibilities included:

  • Evaluation strategy and test scoping

  • Hands-on prototyping and experimentation

  • Analysis of performance, reliability, and limitations

  • Recommendations for responsible AI-assisted design

What was tested?

To evaluate both ends of the design workflow, I focused on two use cases.

High-Fidelity Prototype
Used Figma Make to generate a fully interactive prototype from existing designs, leveraging an Advisor Match flow as the reference.

Prompt-to-POC
Explored how Figma Make could generate alternative proof-of-concepts directly from prompts, using a prior Multi-Order Entry flow as conceptual grounding.

What was out of scope?

These areas were intentionally excluded to keep the evaluation focused on realistic team use cases:

  • Iteration on uploaded wireframes

  • Backend or Supabase integrations

  • Full design system ingestion

Experiment 1: High-Fidelity Prototype from Existing Designs

Goal

Assess whether Figma Make could translate production-ready Figma designs into a near pixel-perfect, fully interactive prototype suitable for usability testing.

Approach

Starting from a previously tested Advisor Match prototype:

  • Imported design frames into Figma Make

  • Refined layouts and behaviors through prompts and targeted div edits

  • Iterated between Figma and Figma Make to improve structure and fidelity

47

Minutes spent building it

1:55

Median load time

23

Total prompts

12

Imports

3

Errors

7

Div edits

Key Findings

The output behaved more like a real website than a traditional Figma prototype

  • Interactive elements such as typeable fields and dynamic states worked well for testing

  • Minor fidelity gaps remained and were not always fixable

  • The process took longer than expected due to iterative trial and tool limitations

Key Findings

The output behaved more like a real website than a traditional Figma prototype

  • Interactive elements such as typeable fields and dynamic states worked well for testing

  • Minor fidelity gaps remained and were not always fixable

  • The process took longer than expected due to iterative trial and tool limitations

Key Findings

The output behaved more like a real website than a traditional Figma prototype

  • Interactive elements such as typeable fields and dynamic states worked well for testing

  • Minor fidelity gaps remained and were not always fixable

  • The process took longer than expected due to iterative trial and tool limitations

Experiment 2: Prompt-to-POC Generation

Goal

Determine whether Figma Make could accelerate early ideation by producing viable interactive concepts directly from prompts.

Approach

Using an older Multi-Order Entry flow as reference:

  • Described the experience in plain language prompts

  • Generated an initial proof-of-concept

  • Iterated through additional prompts to explore alternative solutions

9:42

Minutes spent building it

1:50

Median load time

5

Total prompts

2

Errors

Key Findings

Produced multiple viable workflow options quickly

  • Reduced effort required to stand up early interactive concepts

  • Output favored happy paths and simplification over system depth

  • Risk of shallow UX decision-making increased without close design oversig

Key Findings

Produced multiple viable workflow options quickly

  • Reduced effort required to stand up early interactive concepts

  • Output favored happy paths and simplification over system depth

  • Risk of shallow UX decision-making increased without close design oversig

Key Findings

Produced multiple viable workflow options quickly

  • Reduced effort required to stand up early interactive concepts

  • Output favored happy paths and simplification over system depth

  • Risk of shallow UX decision-making increased without close design oversig

AI Learnings

What Worked Well

Rapid generation of interactive proof-of-concepts

  • Strong inference of expected interactions from existing designs

  • More realistic prototypes for usability testing

  • Faster early communication with product and engineering partners

What Fell Short

Long and inconsistent load times

  • Prototype interactions not retained automatically

  • Frequent homepage reloads during iteration

  • Prompt-first generation encouraged surface-level design thinking

Key Takeaways

Specificity Matters
Clear, detailed prompts produced better results and fewer unintended behaviors.

Naming Is Leverage
Consistent naming conventions improved precision during iteration.

Precision Beats Broad Edits
Targeted div-level changes were more reliable than global prompts.

Structure Improves Translation
Clean Auto Layout and grid usage reduced layout and spacing issues.

Upfront Work Improves Stability
Supplying a PRD and clear flows gives Figma Make stronger context, which reduces breakages and speeds up usable output.

Recommendations

Figma Make is best used for fast POCs and interactive prototypes, not complex multi-page workflows. With maturity, it could support research and early exploration, but it still requires close designer oversight to avoid shallow outcomes.

Let’s make something great.

linkedin.com/in/zachandwhite

zachandwhite@gmail.com

Let’s make something great.

linkedin.com/in/zachandwhite

zachandwhite@gmail.com

Let’s make something great.

linkedin.com/in/zachandwhite

zachandwhite@gmail.com