Hey Reader,

I've been thinking a lot lately about what actually separates designers who grow through this AI moment from those who feel swept along by it.

Two things keep coming up in mentoring calls and in the work I've been reading: taste and self-awareness.

The gems this issue land in that space — what AI changes, the psychology behind the products we make, to the metrics that help us understand if the work we're doing is actually moving things forward.
​

In today's email:

• The AI identity crisis: Figma's Chief Design Officer sat down with Dive Club for a wide-ranging conversation about what AI is really changing in design. If you've been wrestling with what it means to keep a design practice that feels like yours, this one is worth 50 minutes.
  ​
• Design psychology: A free online book that goes where most design resources don't — into your own biases, your org's dysfunction, and the mental models your users bring before they ever touch your product.

⭐️ Join my live with Kit I’m joining Kit’s Creator Educator Shiv Chibber for a live workshop on what’s inside my Kit – not just the content you see, but the systems behind it. It's really for folks who are also growing an audience, so may not be your jam but sharing for those of you who may be interested! Save your spot

Design gems of the week

• ​Figma's Big Bets for the Future of AI Design: A thoughtful conversation with Loredana Crisan, Figma's Chief Design Officer, on why "design is having an identity crisis and a renaissance at the same time" — covers taste, AI as expression vs. outsourcing and what it means to design systems not screens.
  ​
• ​Product Design Psychology: A book for designers who want to better understand the behavioural side of product design. It covers how people think, decide, and move through digital experiences
  ​
• ​Design Engineering: A short piece on the role of the design engineer and how it's changed.
  ​
• ​How to Measure UX and Design Impact: A course by Vitaly Friedman on measuring UX in a way that actually helps teams make better decisions. Useful if you want to get more confident connecting design work to outcomes, choosing the right metrics, and explaining impact without reducing everything to a dashboard.
  ​
• ​Why Systems Thinking is Becoming the Most Important UX Skill: A read on why designers need to look beyond individual screens and understand the broader systems their work sits inside.
  ​
• ​Astryx: A new open source design system that's fully customizable and agent ready – looks promising and might be a nice alternative from shadcn.
  ​
• ​What is product craft in the age of AI and design systems: A spicy(?) take on how design systems alone can't raise the quality ceiling. The best design system can't save a bad team from creating poor experiences.

​
Open design roles worth a look

• ​Director, Product Design (Strava) - San Francisco, Hybrid
• ​Senior Product Designer, AI-Native Product (Wealth) - New York, Hybrid
• ​Staff Visual Designer - Design Systems (Mercury) – Remote within US & Canada
• ​Senior Product Designer (Epoch AI) – Remote

A MESSAGE FROM FRAMER The pages you build last, then never come back to Every site has pages that get built last and never revisited. A confirmation page, a thank-you screen, the page someone sees right after they sign up. Framer Agents works inside your Framer site, so you can point it at that page and let it handle the first pass while you keep an eye on the details. You still decide what actually ships. ​​→ ​Try Framer 3.0 with Agents​

AI raised the ceiling, but the fundamentals of UX remain the same

Yesterday I sat in on Vitaly Friedman's live session, Fixing Frustrating Design Patterns In 2026.​

Ninety minutes on problems our industry has technically known how to solve for two decades — ambiguous X buttons, hover menus that break the moment your mouse moves diagonally, disabled states that look identical to enabled ones.

None of it required new technology to fix. It required someone to notice. Which might be the more uncomfortable finding.

What the session was really about

The headline patterns were navigation (menus, hover triangles, infinite scroll), ambiguous controls (X vs. Cancel vs. Reset), and the hidden/disabled/read-only mess most design systems still get wrong.

A few things worth holding onto:

• The triangle pattern for hover menus — tracking cursor position so a submenu stays open while you move diagonally toward it, rather than closing the instant your mouse leaves a straight line. Amazon solved this in 2001. Most of our menus still haven't caught up.
  ​
• Progress bars usually lie, especially with branching flows. A sidebar showing all steps and your current position is more honest — and more useful.
  ​
• The pattern behind almost every fix: give people an explicit label instead of an icon or a color to interpret. "Apply Filters" beats a checkmark. "Cancel Entire Application" beats a bare X.

None of this requires an LLM. It requires someone caring enough to ask what a control actually communicates.

Why this still matters in the AI moment

Microsoft published a piece on redesigning Copilot around the idea that "output is the new UX" — the interface is the plate, the AI's answer is the meal.

A useful reframe for anyone designing generative products.

So, Copilot is your personal chef, and the response itself is something composed with its own design vocabulary — typography, motion, tone, structure — deciding in real time what to emphasize and what to let fade into the background.

The meal isn't just served. It's plated with intent.

Here's the thing, though: every example Vitaly picked apart is a plate problem. If output is the new UX, the plate doesn't disappear. It has to get quieter and more certain, not just less visible.

A generated answer you can't fully predict is already asking for trust. A save button that might mean discard spends that trust before the meal even arrives.

The ceiling on what's technically possible keeps rising, but the floor — the boring, unglamorous craft of making a control mean one clear thing — hasn't moved. If anything it matters more. There's less room left to hide behind "well, at least the AI part is impressive."

If your team is deep in an AI feature redesign right now, ask the boring question too: does the X button mean "close" or "discard"? You'll probably find out you don't know.

Have a topic you'd like me to write about? Reply to this email and let me know! I read every reply.