CHALLENGE THE PROBLEM
Before moving into wireframes or UI, I used AI as a critique partner to pressure-test my initial problem framing.
Rather than asking AI to solve the problem for me, I used it to expose assumptions I might otherwise carry into the design process.
I asked questions such as:
• What am I assuming about users?
• Whose workflow is missing?
• What business constraint changes the problem?
• Am I solving the root issue?
This gave me an additional layer of critique early in the process, when changing direction was still inexpensive.
AUTOMATE THE BUSYWORK
I used AI-assisted tools within my Figma workflow to accelerate repetitive parts of exploration and prototyping.
This included
• Generate alternate UI directions
• Create realistic placeholder content
• Explore empty, error, and edge states
• Build clickable prototypes faster
The goal was not to accept the first AI-generated solution.
It was to create more viable options faster so I could spend more time evaluating hierarchy, usability, interaction patterns, and business fit.
USE DEV MODE AS TRUTH
Once the interaction model and UI direction were validated, I used Figma Dev Mode to reduce ambiguity between design and development.
Engineers could directly inspect:
• Spacing and sizing
• Typography and color values
• Component variants and interaction states
• Assets and export settings
• Layout behavior and implementation context
This reduced the need for separate specification documents and made design intent easier to reference throughout implementation.
USER + BUSINESS CONTEXT
The constraints still
came from people.
I still rely on research, stakeholder conversations, technical collaboration, and business context to understand what users are trying to accomplish, where friction exists, and what success means for the business.
AI helps me explore the solution space, but it does not define the problem.
DESIGN OWNERSHIP
I make the
final design decision.
I review AI output for unsupported assumptions, inconsistent interaction patterns, accessibility issues, missing edge cases, technical feasibility, and alignment with user and business needs.
AI accelerates my workflow, but accountability for the final product remains with me.
RESEARCH & VALIDATION
Real users still
validate the work.
I still rely on interviews, usability testing, stakeholder feedback, product data, and observation to understand whether the workflow matches real behavior and whether the solution improves the intended outcome.
AI help me prepare for validation, people provide the evidence.
I use AI where it creates leverage: exploration, repetitive production work, content generation, and early critique.I keep the parts that require judgment firmly human: understanding users, balancing business and technical constraints, validating assumptions, and deciding what should ultimately ship.
FASTER
EXPLORATION
More viable directions
before committing.
EARLIER
CRITIQUE
Biases and missing assumptions
surfaced sooner.
CLEARER
HANDOFF
Less interpretation between
design and build.