Figma's AI Tools: Design Automation or Creative Assistant?

"Figma's AI just redesigned my button," my designer colleague messaged me.

"Did it make it better?" I asked.

"...Different. More generic. Technically correct. Totally soulless."

This exchange captures the tension in Figma's AI features, launched in June 2024 and refined throughout the year.

The tools work. They generate layouts, match design systems, write copy, and suggest improvements.

The question isn't whether they work—it's whether they should.

After using Figma's AI features for six months across multiple projects, I've seen both brilliant automation and creative flattening.

Here's what Figma's AI actually does, what works, and where it misses the point entirely.

What Figma Added: The AI Feature Set

Launched June 2024, expanded through October:

1. AI-Powered Auto-Layout

What it does:

Select elements → Figma suggests auto-layout structure → Apply in one click

The promise: No more manually configuring spacing, padding, constraints.

Real example:

I have three cards in varying layouts (different padding, inconsistent spacing).

Figma AI:

• Detects they're similar elements
• Suggests unified auto-layout
• Standardizes spacing (16px padding, 12px gap)
• Makes responsive (proper constraints for different screen sizes)

Time saved: What took 10 minutes of manual adjustment takes 10 seconds.

2. Design System Matching

What it does:

Analyzes your existing designs → Suggests components that should use design system styles

Real example:

I've used three different shades of blue across a design (oversight, not intention).

Figma AI:

• Flags inconsistency
• Suggests: "These look like they should use primary-blue from your design system"
• Offers to replace all instances

Value: Catches inconsistencies I missed.

3. Content Generation

What it does:

Generates placeholder text, images, or realistic content based on context

Real example:

Designing a user profile card.

Figma AI generates:

• Realistic name ("Sarah Chen")
• Appropriate bio ("Product designer based in London, passionate about accessible design")
• Profile photo (AI-generated, matches context)

Better than "Lorem ipsum" or manual placeholder creation.

4. Layout Suggestions

What it does:

Analyzes content → Suggests alternative layouts

Real example:

I have a feature comparison table (5 columns, 12 rows).

Figma AI suggests:

• Alternative 1: Card-based layout (one card per feature)
• Alternative 2: Tabbed interface (group features by category)
• Alternative 3: Accordion (collapse/expand sections)

Each comes with working prototype.

5. Accessibility Checks

What it does:

Flags accessibility issues (contrast, text size, hit targets)

Real example:

My text is #6B6B6B on #FFFFFF background.

Figma AI:

• Flags: "Contrast ratio 4.2:1. Fails WCAG AA for normal text (requires 4.5:1)"
• Suggests: "Use #5A5A5A for 4.6:1 ratio"

Immediately useful.

What Works Brilliantly

Use Case 1: Enforcing Design System Consistency

Scenario: Large team, shared design system, inevitable drift

The problem:

• Designer A uses primary-button component
• Designer B doesn't know it exists, creates similar button from scratch
• Now we have two slightly-different primary buttons
• Multiply by 10 designers × 6 months = chaos

Figma AI solution:

• Scans all designs
• Flags components that should reference design system but don't
• Offers bulk replacement

Result: I cleaned up 6 months of design system drift in 2 hours.

This is transformative for design ops.

Use Case 2: Responsive Design Tedium

Scenario: Design for desktop → now make mobile version

The tedious work:

• Adjust every spacing value
• Rearrange elements for narrow screen
• Fix text wrapping
• Adjust image sizes
• Test breakpoints

Figma AI approach:

• Analyzes desktop design
• Suggests mobile layout automatically
• Adjusts spacing proportionally
• Reflows content intelligently

Accuracy: ~70% (requires manual refinement, but gets you 70% there)

Time savings: Massive. What took 3 hours takes 45 minutes.

Use Case 3: Prototyping Speed

Scenario: Need to show stakeholder three different layout options

Old workflow:

• Design option A manually
• Duplicate
• Redesign as option B manually
• Duplicate
• Redesign as option C manually
• Time: 6+ hours

With Figma AI:

• Design option A
• Ask AI: "Show me alternative layouts for this content"
• Review AI suggestions (3-5 alternatives generated)
• Refine chosen alternatives
• Time: 2 hours

4× faster for exploration.

What Feels Deeply Wrong

Problem 1: The Generic Design Aesthetic

Observation: AI-suggested layouts all look... similar.

They look like:

• Rounded corners (8px radius)
• Sans-serif fonts (Inter, probably)
• Muted colors (blues, grays, whites)
• Generous whitespace
• Card-based layouts

This is SaaS Design Starter Pack.

Why this happens:

AI is trained on existing designs. Most modern designs follow conventions. AI optimizes for "what designs typically look like."

Result: Everything trends toward average.

Example:

I was designing a playful consumer app. Bright colors, unconventional layouts, personality.

Figma AI suggestions:

• Soften the colors (too bright)
• Increase whitespace (too dense)
• Standardize spacing (inconsistent feels unprofessional)

Every suggestion pushed toward generic.

I ignored them. Design stayed playful. Stakeholders loved it.

If I'd accepted AI suggestions, design would be "fine" and completely forgettable.

Problem 2: AI Doesn't Understand Intent

Scenario: Design element is intentionally unusual

Example:

I made a CTA button small and subtle (the page goal was reading content, not clicking button).

Figma AI:

• Flags: "Button is below minimum recommended size (44×44px touch target)"
• Suggests: Increase size to meet accessibility standards

Technically correct. Strategically wrong.

The small button was intentional design choice. AI can't distinguish between:

• Mistake: Forgot to make button big enough
• Intent: Button should be subtle

AI assumes mistake.

Problem 3: Loss of Design Cohesion

Good design has internal logic:

• Elements relate to each other
• Spacing has rhythm
• Colors tell a story

AI suggestions optimize locally, not globally.

Example:

My design used 16px as base spacing unit. Everything was multiples of 16 (16, 32, 48, 64).

This creates visual rhythm.

Figma AI suggested:

• Change this padding from 48px to 40px (better proportion for this specific element)

Locally correct. 40px might look better in isolation.

Globally destroys rhythm. Now spacing is inconsistent (16, 32, 40, 48, 64).

AI doesn't see the system-level pattern.

Problem 4: Accessibility Theatre

AI flags accessibility issues. This is good.

But it can't assess whether design is actually accessible to real users.

Example:

Figma AI approved:

• Proper contrast ratios ✓
• Sufficient text size ✓
• Adequate touch targets ✓
• Color isn't only differentiator ✓

But the layout was cognitively overwhelming:

• 12 different actions on one screen
• No clear visual hierarchy
• Important and unimportant elements given equal weight

Technically accessible. Functionally unusable.

AI checks rules. Can't evaluate experience.

Designer Reactions: The Split

I surveyed 30 designers about Figma AI (informal, not scientific):

Group 1: "It's Amazing" (~40%)

Profile:

• Early-career designers
• Working on design systems
• High-volume production work
• Agency/consultancy (lots of projects, tight deadlines)

What they value:

• Speed (crank out more designs faster)
• Consistency (design system enforcement)
• Learning (AI teaches best practices)

Quote:

"Figma AI is like having a senior designer checking my work. It catches mistakes I'd miss and suggests improvements I wouldn't have thought of."

Group 2: "It's Dangerous" (~35%)

Profile:

• Senior designers
• Brand/creative work
• Custom/unconventional designs
• Prioritize originality over speed

What they fear:

• Homogenization (everything looks the same)
• Deskilling (junior designers won't learn fundamentals)
• Creativity loss (AI nudges toward safe, conventional choices)

Quote:

"AI is training a generation of designers to make everything look like a B2B SaaS dashboard. Where's the personality? The risk-taking? The weird ideas that sometimes work brilliantly?"

Group 3: "It's a Tool" (~25%)

Profile:

• Mid-senior designers
• Pragmatic about AI
• Use features selectively

Their approach:

• Use AI for tedious work (spacing, consistency checks)
• Ignore AI for creative decisions
• Treat suggestions as starting point, not answer

Quote:

"I use AI like I use templates. Helpful for scaffolding. Terrible if you stop there."

When to Use Figma AI (And When to Ignore It)

Use AI For:

1. Design system auditing

Finding inconsistencies across large design files.

AI is better than humans at systematic pattern detection.

2. Responsive adaptation

Creating mobile/tablet versions of desktop designs.

AI handles tedious spacing adjustments.

3. Accessibility compliance

Checking contrast, sizing, touch targets.

AI catches technical violations.

4. Rapid prototyping

Generating layout alternatives quickly.

AI accelerates exploration phase.

5. Content generation

Placeholder text, images for prototypes.

AI creates more realistic placeholders than "Lorem ipsum."

Ignore AI For:

1. Brand-defining decisions

Color palettes, typography choices, overall aesthetic.

This requires human judgment about brand positioning.

2. Unconventional layouts

When you're intentionally breaking rules.

AI will push you toward convention.

3. Emotional design

When design needs to evoke specific feeling.

AI optimizes for "acceptable," not "emotionally resonant."

4. Strategic trade-offs

When design serves specific goal that requires trade-off.

(Example: Small button because you don't want clicks)

5. Creative exploration

When you're exploring weird ideas.

AI suggestions kill weird. Weird is often where innovation lives.

The Bigger Question: What Is Design For?

Figma's AI forces a philosophical question:

Is design problem-solving or creative expression?

If Design = Problem-Solving

Goal: Create functional, usable, accessible interfaces

AI is great.

It solves problems efficiently. Enforces best practices. Reduces errors.

Outcome: High-quality, professional designs produced faster.

Trade-off: Less differentiation. Everything trending toward similar solutions.

If Design = Creative Expression

Goal: Create distinctive, memorable, emotionally resonant experiences

AI is limiting.

It optimizes toward average. Discourages risk. Flattens personality.

Outcome: Unique designs, brand differentiation.

Trade-off: Slower, more expensive, higher risk of accessibility/usability issues.

The Truth: Design Is Both

Most work requires both problem-solving and creativity.

The challenge: Knowing when to prioritize which.

Internal admin dashboard? Problem-solving. Use AI. Get it done efficiently.

Consumer brand landing page? Creative expression. AI is starting point at best.

Figma's AI doesn't help you make this distinction.

It treats all design work as problem-solving to be optimized.

What's Next: Predictions

Prediction 1: AI-Generated Design Systems

Current: AI helps you maintain design system

Future: AI generates entire design system from brand inputs

Input: Brand colors, fonts, positioning

Output: Complete design system (components, spacing scales, color palettes, type scales)

Timeline: 12-18 months

Prediction 2: Conversational Design

Current: Use tools to create designs

Future: Describe what you want, AI designs it

"Create a pricing comparison table. Three tiers. Emphasize middle tier. Use our design system."

AI generates design matching description.

Timeline: 18-24 months

Prediction 3: Real-Time User Testing Integration

Current: AI suggests designs based on convention

Future: AI suggests designs based on actual user testing data

"This layout performed 23% better in A/B tests for similar products."

Requires integration with analytics tools.

Timeline: 2-3 years

Prediction 4: Designer Roles Split

Current: Designers do everything (research, ideation, execution, refinement)

Future: Roles split:

• Design strategists: High-level creative direction, brand work (AI can't do this)
• Design producers: Execution, production, system maintenance (AI augments this heavily)

Similar to photography:

• Art directors (creative vision)
• Photo editors (execution, refinement)

Timeline: Already happening

---

TL;DR: Figma AI six months in

What Figma added (June 2024):

• AI auto-layout
• Design system matching
• Content generation
• Layout suggestions
• Accessibility checks

What works brilliantly:

• Design system consistency enforcement
• Responsive design adaptation (70% automated)
• Prototyping speed (4× faster exploration)
• Accessibility compliance checking

What feels wrong:

• Pushes everything toward generic SaaS aesthetic
• Can't understand design intent
• Optimizes locally, breaks global cohesion
• Checks accessibility rules but not actual usability

Designer reactions:

• 40%: "It's amazing" (early-career, high-volume work)
• 35%: "It's dangerous" (senior, creative work)
• 25%: "It's a tool" (pragmatists using selectively)

When to use AI:

• Design system audits
• Responsive adaptation
• Accessibility checks
• Rapid prototyping
• Placeholder content

When to ignore AI:

• Brand-defining decisions
• Unconventional layouts
• Emotional design
• Strategic trade-offs
• Creative exploration

The core tension:

Design = problem-solving (AI helps) + creative expression (AI limits)

Knowing which is needed when = still human judgment

Chaos integrates design thinking into task management—organize your work with the same thoughtfulness you bring to design. Start your free 14-day trial.