Design in Everything - by Thomas Walichiewicz

I see a ton of nothingburger articles about AI and design every day, so I figured I’d throw my two cents into the void.

Lately, I’ve been deep in the weeds building enterprise systems to automate complicated, laborious, mission-critical work:

AI/ML tooling pipelines
Product instance generators
Standardized installers across 60+ enterprise tools

(Sexy stuff, I know.)

At their core, each of these projects solved a human-level problem: eliminate waste and free people from repetitive tasks so they can focus on decisions that actually matter.

In that same spirit, I turned the lens on myself—and put together real-life release notes for the latest version of me:

Product Designer 2.5 - Release Notes
UpdatedI explore more broadly	I use AI to spin up multiple design directions and pressure-test concepts early—less about "more options," more about sharpening judgment on why one direction wins.
UpdatedI automate repetitive work	Instead of hand-coding elements or rewriting copy for every variant, I script reusable components and use LLMs to translate content across use cases. That lets me stay focused on systems, not scaffolding.
UpdatedI get to clarity faster	AI helps me burn through early noise so I hit friction points sooner. That's where meaningful design decisions actually begin.
NewI redefine the problem space and embed earlier	With less time spent in production, I go upstream—reframing what should even exist. I dig into overlooked workflows, unmet needs, and the ripple effects of product decisions. The best design isn't always a better UI. Sometimes it's changing the brief entirely.
NewI design for second- and third-order effects	I think more about scale, team ops, and behavior. What happens when this pattern gets reused 10,000 times? What kind of culture are we shaping? It's not just usability anymore—it's long-term coherence and trust.
NewI shape vision through prototypes, not slides	Fully interactive prototypes in hours. I use them to pitch bolder, more abstract ideas—the kind that usually get dismissed for being "too big" or "too early." Now I can show, not just tell.
NewI create reusable thinking, not just reusable components	Beyond design systems, I document principles, tradeoffs, and mental models—so others (not just designers) can make smarter decisions without reinventing the wheel.

Real AI Workflows in Practice

Here are specific ways I’m using AI tools today:

Rapid Concept Validation

Tool: Claude + Figma
Workflow: Feed existing designs into Claude → Generate 5 conceptual variations → Import as wireframes → Test with users in <2 hours
Impact: What used to take 2 days now takes 2 hours

Component Documentation

Tool: GitHub Copilot + Custom Scripts
Workflow: Write component specs in natural language → Auto-generate JSDoc comments → Create usage examples → Update design system docs
Impact: 80% reduction in documentation time

User Research Synthesis

Tool: GPT-4 + Notion
Workflow: Transcribe interviews → Extract themes → Map to existing personas → Generate insight reports with quotes
Impact: Turn 20 hours of interviews into actionable insights in 30 minutes

Design QA Automation

Tool: Custom Python scripts + Vision APIs
Workflow: Screenshot implementations → Compare to design files → Flag spacing/color/typography mismatches → Generate fix tickets
Impact: Catch 95% of visual bugs before they ship

Accessibility Testing

Tool: AI-powered browser extensions
Workflow: Run designs through contrast checkers → Generate ARIA labels → Suggest keyboard navigation improvements
Impact: Ship more accessible products without manual audits

Key Takeaways

AI amplifies judgment, not just output - The value isn’t in generating more options, but in reaching meaningful design decisions faster
Automation creates space for strategy - Less time on production means more time understanding second-order effects and long-term impact
Show, don’t tell becomes scalable - Interactive prototypes in hours means bolder ideas get real consideration instead of dying in committee