Jenny Wen (head of design at Claude)01 March 2026

The design process is dead. Here’s what’s replacing it.

7Frameworks

15Insights

Episode overview

Jenny Wen argues that the traditional sequential design process is effectively dead because AI-assisted engineering now produces working software too quickly for designers to lead primarily through polished mocks. She describes a role split between short-horizon vision-setting and hands-on execution: pairing with engineers, prototyping in code, polishing implementations, and learning from real users. The conversation also explores trust through rapid iteration, the continuing value of human accountability and judgment, changing management practices, and three desirable hiring archetypes: strong generalists, deep specialists, and adaptable early-career designers.

Key ideas

AI-accelerated engineering is forcing design teams to abandon rigid research-to-mock-to-handoff workflows.
Design work is stratifying into implementation support and short-horizon vision-setting, often through working prototypes rather than polished decks.
Jenny's time spent mocking and prototyping has fallen from roughly 60–70% to 30–40%, while pairing with engineers and implementation have grown.
Non-deterministic AI products need to be tested with real models, data, users, and use cases rather than fully represented in static mocks.
Teams can build trust through speed by labeling early releases honestly, responding to feedback, and visibly shipping improvements.
AI may develop stronger taste and judgment, but humans remain accountable for deciding what matters and what gets built.
Effective design managers should stay close to the craft, combine team direction with people management, and sometimes choose symbolically important low-leverage work.
Jenny highlights strong generalists, deep specialists, and highly adaptable early-career designers as valuable hiring archetypes.

Transcript available · source text is retained privately and is not published

Frameworks in this episode

Self-Mastery3 steps

AI Introspection Rubric

Point an AI agent at a folder of your own notes to surface criteria you hold implicitly and turn them into a rubric.

Read →

Strategy3 steps

Building Trust Through Speed

Ship an honest 'research preview' early, then visibly iterate on feedback so speed builds trust instead of eroding it.

Read →

Leadership3 steps

Deliberate Low-Leverage Leadership

Choose the nitty-gritty tasks 'anyone could do' — because when a senior leader does them, they become high leverage.

Read →

Innovation3 steps

Legibility Framework for Spotting Frontier Ideas

Hunt for illegible ideas — the ones with real energy that nobody can quite articulate yet — and translate them.

Read →

Leadership3 steps

Roasting Plus High Standards

Let the team feel safe enough to roast you, then apply high standards without fear — safety makes rigor easier.

Read →

Productivity3 steps

Stratified Design Work

Split your time between supporting engineers' execution and setting a 3-6 month vision, not polishing mocks.

Read →

Leadership4 steps

Three Designer Hiring Archetypes

Hire for one of three profiles — block-shaped generalist, deep-T specialist, or the crack new grad.

Read →

People & resources mentioned

Attributed to the moment in the episode. Timestamps are approximate.

People · 11

Jenny WenMentionsJenny was head of design for Claude, is now leading design for Claude Co-work. Prior to that, she was director of design at Figma
“Today's guest is Jenny Wen. Jenny was head of design for Claude, is now leading design for Claude Co-work.”
▶ Watch at ~1:00approx.
Emily Lin HashamMentionsmentioned
“A huge thank you to no 111 and Emily Lin Hasham for suggesting topics and questions for this conversation.”
▶ Watch at ~1:30approx.
Marc AndreessenMentionsMark Andre had this visual on some podcasts have just imagined a 100,000 drones just coming out of China just at us.
“we had Mark Andrees on the podcast. We kind of called it the F Fshape or E shape”
▶ Watch at ~50:00approx.
Mike KriegerMentionsMike is chief product officer at Anthropic the company behind Claude. He's also the co-founder of Instagram.
“I think I've actually seen like Mike Kger before like put put in PRs himself.”
▶ Watch at ~56:30approx.
Evan TanaCoineda partner at SPC
“it was Evan Tana who is a partner at SPC”
▶ Watch at ~1:02:00approx.
Robert CaroCoinedPower Broker by Robert Caro
“The first one is The Power Broker by Robert Caro”
▶ Watch at ~1:09:00approx.
Robert MosesMentionsmentioned
“it's nice to sort of like read a really nuanced view of somebody Robert Moses”
▶ Watch at ~1:09:30approx.
Bill HayesCoinedthe partner of the scientist Oliver Sacks around the time that Oliver Sacks died
“a book called Insomniac City, which is written by Bill Hayes”
▶ Watch at ~1:10:00approx.
Oliver SacksMentionsscientist
“Bill Hayes, who is was the partner of the scientist Oliver Socks around the time that Oliver Sox died”
▶ Watch at ~1:10:00approx.
Lenny RachitskyMentionstoday we've got another very special compilation episode something I've been pulling on more and more with the podcast and the newsletter
“this has like very little to do with the stuff on your podcast, Lenny, but it's just like it's just a book”
▶ Watch at ~1:10:30approx.
Noah WyleMentionsmentioned
“I only recently realized Noah Wy was in ER as like a younger person and now he's like the head of this.”
▶ Watch at ~1:12:00approx.

Resources · 29

FigJamMentionssoftware · Figma, Inc.
“where she led the design teams behind Fig Jam and Slides.”
▶ Watch at ~1:30approx.
SquareMentionscompany · Block, Inc.
“She was also designer at Dropbox and Square and Shopify.”
▶ Watch at ~1:30approx.
DropboxMentionscompany · Drew Houston and Arash Ferdowsi
“She was also designer at Dropbox and Square and Shopify.”
▶ Watch at ~1:30approx.
Figma SlidesMentionssoftware · Figma
“where she led the design teams behind Fig Jam and Slides.”
▶ Watch at ~1:30approx.
ShopifyMentionscompany · Shopify
“She was also designer at Dropbox and Square and Shopify.”
▶ Watch at ~1:30approx.
OrkesMentionscompany
“This episode is brought to you by Orcus, the company behind Open-source Conductor”
▶ Watch at ~3:00approx.
ConductorMentionssoftware · Orkes
“the company behind Open-source Conductor, the orchestration platform powering modern enterprise applications.”
▶ Watch at ~3:00approx.
v0Usessoftware · Vercel
“we are using tools like Claude Code and and bzero and whatnot to to start to spin up prototypes”
▶ Watch at ~10:30approx.
AnthropicMentionscompany
“A good amount of time at Anthropic is actually just like catching up on what people what's happening at the company.”
▶ Watch at ~13:00approx.
Visual Studio CodeUsessoftware · Microsoft
“I use it mostly in the with with VS Code in the IDE because I'm usually tweaking frontend stuff”
▶ Watch at ~19:00approx.
Claude CoworkUsessoftware · Anthropic
“I've basically shifted all of my chat use cases over to co-work”
▶ Watch at ~19:00approx.
FigmaUsessoftware · Figma
“I am still using Figma. Yes. Yes.”
▶ Watch at ~20:00approx.
TelegramMentionssoftware · Telegram
“another way to chat with it through WhatsApp and telegram and SMS”
▶ Watch at ~33:30approx.
OpenClawMentionssoftware · Peter Steinberger
“with openclaw claude mbot all the names the innovation one of the big innovations is like another way to chat with it”
▶ Watch at ~33:30approx.
WhatsAppMentionssoftware · Meta
“another way to chat with it through WhatsApp and telegram and SMS”
▶ Watch at ~33:30approx.
OmniMentionssoftware · Omni
“Omni takes a different approach. They have a semantic layer built in”
▶ Watch at ~34:30approx.
SocraticaRecommendsevent
“my one call here too is like I went to a school that started something called Socratica”
▶ Watch at ~51:30approx.
Radical CandorMentionsbook · Kim Scott
“reminds me of what was it candid radical cander just this combination of caring deeply and challenging directly.”
▶ Watch at ~1:01:30approx.
Legibility FrameworkMentionsother · Evan Tana
“I hear you're a big fan of something called the legibility framework.”
▶ Watch at ~1:01:30approx.
Claude StudioMentionssoftware · Anthropic
“there was this internal prototype that we've called like cloud studio”
▶ Watch at ~1:04:00approx.
The Power BrokerRecommendsbook · Robert Caro
“The first one is The Power Broker by Robert Caro, which is an which is an incredibly aggressive recommendation”
▶ Watch at ~1:09:00approx.
The Worst Person in the WorldMentionsfilm · Joachim Trier
“it's a Norwegian film by the same director who did The Worst Person in the World.”
▶ Watch at ~1:11:00approx.
Sentimental ValueRecommendsfilm · Joachim Trier
“I watched A Sentimental Value recently. I watched it on a plane”
▶ Watch at ~1:11:00approx.
The PittRecommendstv show
“and then I would also recommend obviously The Pit season 2.”
▶ Watch at ~1:11:30approx.
ERMentionstv show
“I only recently realized Noah Wy was in ER as like a younger person”
▶ Watch at ~1:12:00approx.
RetroRecommendssoftware · Nathan Sharp and Ryan Olson
“for folks who don't know about Retro, it's sort of like this small community photo sharing app”
▶ Watch at ~1:12:30approx.
iA WriterUsessoftware · Information Architects
“I have this this folder basically of like local IIA writer for and I basically just like write whatever”
▶ Watch at ~1:14:30approx.
Twitter/XUseswebsite
“I'm on Twitterx is what we're calling it these days. Um it's jenny_wen.”
▶ Watch at ~1:16:30approx.
Lenny's PodcastMentionspodcast · Lenny Rachitsky
“You can find all past episodes or learn more about the show at lennispodcast.com.”
▶ Watch at ~1:17:00approx.

Spot an error or want something removed? Request a correction or removal.

Insights & moments

The myth-busts, hot takes, explainers, and tools worth keeping.

Myth Buster· 2

Myth Buster20:00

Why Figma Still Matters in a Code-First World

Against the Twitter debate that code has killed the need for Figma, Jenny says Figma still fills a real gap. Its strength is exploring 8-10 different options at once, plus fine visual and interaction details like typography. Coding tools are too linear for that because you invest heavily in one direction and iterate on it.

Best design comes from throwing many ideas at the wall and curating
Figma excels at exploring 8-10 different directions, including micro visual details
Coding tools are super linear, so you over-invest in one direction
Figma is still the right tool for exploring typography and interaction details

“I think the best design happens when you're able to just like throw a bunch of ideas at the wall and curate”
Jenny Wen · 20:30

“Right now, coding or or right now working with some of these coding tools doesn't lend itself super well to that because it's super linear.”
Jenny Wen · 21:00

#figma#design-tools#exploration

Myth Buster44:30

Co-work Wasn't Really Built in 10 Days

The viral claim that Claude Co-work was built in 10 days is misleading. Jenny explains the 10 days was only the sprint to get it from an internal state to something shippable. Many people had prototyped its pieces over a long journey, and the idea kept coming back until it suddenly felt like the right moment.

The 10 days was only getting it from an internal version to something ready to ship externally
Many form factors (to-do lists, multiple-choice questions) were prototyped internally beforehand
The '10 days' became the viral takeaway but understates the long journey and many people involved
The idea kept coming back until it suddenly felt obvious and was the right moment

“for some reason that became like the viral thing that got taken away from all of the the sort of co-work announcements is that it…”
Jenny Wen · 44:30

“it feels like it was so obvious all along but there was a long long journey to get there.”
Jenny Wen · 45:00

#co-work#product#shipping#prototyping

Hot Take· 3

Hot Take05:00

The Classic Design Process Is Basically Dead

Jenny gave a Berlin talk called 'don't trust the design process,' arguing the taught research-then-diverge-converge ritual designers treated as gospel is over. It was dying before AI, but engineers now spinning up features on their own kill it faster. Designers have to let go of that process.

The traditional research/diverge-converge process was treated as gospel and fiercely preserved
It was already dying before AI arrived
Engineers spinning up 'seven clouds' of features accelerate its death
Designers now have to let go of the old process entirely

“it's like this process that we sort of treated as gospel and and tried so hard to preserve and we were like trust the process.…”
Jenny Wen · 05:30

“I think as as designers, we really have to let go of that process. And I think that's the big thing that's changing.”
Jenny Wen · 06:00

#design-process#ai#workflow

Hot Take28:00

AI Will Get Better at Taste, But Someone Must Be Accountable

Asked where human brains stay valuable, Jenny predicts AI's sense of taste and judgment will keep improving and that designers may be clinging to taste too much. But she argues the hard parts of building software are often not the building, they're the disputes over what should go in, and someone still has to be accountable for the decision, like an engineer signing off on AI-written code.

AI will get better at taste and judgment; designers may be overvaluing their edge there
The hard parts of building software are often not building it but deciding what goes in
AI can weigh in on disputes but can't resolve a disagreement between two people
Someone still needs to be accountable for the decision, like an engineer signing off on code

“a lot of the hard parts of building software are actually like not building it.”
Jenny Wen · 29:00

“someone still needs to be accountable for the decision you know the same way that like even though cloud can write all this code for…”
Jenny Wen · 31:00

#ai#taste#judgment#accountability

Hot Take54:30

Low-Leverage Time Is Actually High Leverage

Jenny pushes back on the standard management advice to drop 'low-leverage' tasks. The best leaders she respects deliberately take on seemingly low-leverage work, dogfooding the product, repro-ing bugs, even filing PRs, and that becomes high leverage precisely because it's them doing it. It builds product familiarity and signals that nothing is beneath them.

Standard advice: sort work into what only you can do vs. what anyone can do, and drop the rest
The best leaders instead take on low-leverage tasks themselves, making them high leverage
Examples: senior leaders dogfooding, repro-ing bugs, sharing logs, even filing PRs
It builds deep product familiarity and signals nothing is below the leader

“some of the their best traits is that they choose like lowlever tasks that they take on themselves and that actually ends up being actually…”
Jenny Wen · 55:30

#management#leadership#leverage

Explainer· 4

Explainer06:30

Design Work Has Stratified Into Two Types

Jenny says design work now splits into two kinds. The first is supporting implementation and execution, where engineers make scrappy versions fast and designers no longer have time to make beautiful mocks. The second is setting vision and direction, but that vision is now 3-6 months out instead of 2-10 years, and is often just a prototype rather than a beautiful deck.

Type one: supporting implementation, where engineers ship scrappy versions and designers lack time for beautiful mocks
Type two: creating vision and direction to point the team at one shared cause
Vision horizons shrank from 2/5/10 years to roughly 3-6 months because too much is changing
A vision is now often just a prototype that points people in the right direction, not a beautifully story-told deck

“you as a designer actually like do not have the time to make these beautiful mocks anymore or to like kind of lead in this…”
Jenny Wen · 07:00

“we don't know what's going to happen in 2 years. There's too much changing and it usually becomes a vision that's like 3 to 6…”
Jenny Wen · 07:30

#design#vision#execution#prototyping

Explainer11:30

You Can't Mock Up Non-Deterministic AI

Jenny argues that because AI models are non-deterministic, you literally cannot mock up all the states or even build a clickable prototype. You have to use the actual models and watch real people discover use cases you didn't design for, which is why shipping and learning beats theorizing.

AI models are non-deterministic, so you can't mock up all the states
You can't even make a clickable prototype with the real behavior
You have to design for use cases but discover them as people actually use the product
Executing with real data and real users leads to a better product

“You can't you just can't you can't mock up all the states, you know, and you can't theorize and you can't even make uh like…”
Jenny Wen · 12:00

#ai#prototyping#product#non-deterministic

Explainer24:30

Shipping Early Without Losing Trust

Jenny explains how to ship rough products, like Claude Co-work's 'research preview' label, without damaging the brand. The key is making and keeping a promise: put it out knowing it's flawed, then visibly iterate on user feedback. You lose trust only if you release something early and then nothing ever happens.

Label early products as research previews and caveat the flaws up front
Ship when the benefits clearly outweigh the cons even if quality is rough
Make a promise to users that you'll take feedback and keep improving, and show it publicly
You degrade the brand only if you release early and then nothing ever happens

“the way that you really lose trust around quality and releasing something early is if you release it early and then nothing ever happens.”
Jenny Wen · 26:00

“It's yeah, it's building trust through speed, but also just like making people feel like they've been heard”
Jenny Wen · 27:00

#product#trust#shipping#brand

Explainer46:30

The Three Designer Archetypes Jenny Hires For

Jenny names three archetypes she finds most interesting to hire right now. Strong generalists who are 'block-shaped' (80th percentile in several core skills), deep specialists whose T goes far deeper than most (top ~10%), and the overlooked 'craft new grad' who is early-career, humble, eager, and unburdened by baked-in processes.

Strong generalists: 'block-shaped', ~80th percentile across several core skills, easy to flex as roles stretch
Deep specialists: T-shaped but the tip goes far deeper, roughly top 10% in the industry
Craft new grad: early-career, wise beyond their years, humble, eager, and a quick learner
The new grad is valuable precisely for a blank slate with no baked-in processes or rituals
Above all she looks for resilience and willingness to adapt to new tools and methods

“there's probably three archetypes of folks that are really interesting to me right now.”
Jenny Wen · 46:30

“having somebody who almost has like a blank slate and is just like a really quick learner and is really eager to learn um new…”
Jenny Wen · 49:00

#hiring#design#talent#archetypes

Story· 1

Story13:00

The Best AI News Is Inside the Lab's Slack

Jenny describes that a big part of her day at Anthropic is just keeping up with what's happening internally, across research and labs teams prototyping under code names. She calls the internal Slack a gold mine and says the best AI news is actually inside these companies, not the public feeds.

A good chunk of the job is catching up on internal model developments and prototypes
Teams run many projects under code names she has to track to spot what's coming
The internal Slack is a 'gold mine' she's excited to read
The best AI news is internal, not the public feeds everyone else watches

“a lot of it is like, yeah, our Slack is a gold mine, you know, like I'm just excited to to read through the things…”
Jenny Wen · 15:00

“I think that is the the best AI news is probably internally if you're ever at one of these companies in the slack.”
Jenny Wen · 15:30

#anthropic#culture#workflow#ai

Q&A· 1

Q&A53:30

Would You Hire Claude as a Designer?

Asked whether Claude is good enough to hire as a designer, Jenny says not yet. It isn't a strong generalist, deep specialist, or craft new grad. It's pretty good at a first pass and at presenting many different ideas, but nothing feels special or hirable yet, which she notes is good news for designers for now.

Claude is not yet at the level of a designer you'd hire
It isn't any of the three hireable archetypes yet
It's good at a first pass and at presenting many different ideas
Nothing feels special or hirable yet, though it's improved a lot in the last year

“I think it's it's pretty good at a first pass and at presenting a bunch of different ideas to you, but uh nothing there quite…”
Jenny Wen · 54:00

#claude#ai#design#hiring

Tool· 1

Tool18:30

Jenny's AI Stack: A Fully Claude House

Jenny walks through the tools she uses as a designer: Claude chat, increasingly Claude Co-work for longer-running tasks, and Claude Code inside VS Code for frontend tweaks. She's also using Claude Code remotely via mobile and Slack, where a quick at-mention produces a finished PR.

Shifted most chat use cases to Claude Co-work for longer-running tasks
Uses Claude Code in VS Code because she tweaks frontend and wants to see the code
Uses Claude Code remotely via mobile and Slack; at-mention Claude and pick up a finished PR
Describes the team as 'a fully Claude house'

“It's really fun for somebody to say like oh yeah this this one icons off or something and you just at mention Claude and Claude…”
Jenny Wen · 19:30

“I think we're a fully Claude house here.”
Jenny Wen · 19:30

#tools#claude#ai-stack#design

Takeaway· 3

Takeaway17:30

How a Designer's Time Split Has Changed

Jenny gives a concrete before/after of how her time is spent. A few years ago mocking and prototyping was 60-70% of the work. Now mocking is 30-40%, another 30-40% is jamming and pairing directly with engineers, and a growing slice is actually building and shipping code.

A few years ago: 60-70% mocking/prototyping, ~20% jamming with engineers, ~10% coordination
Now: mocking is only 30-40% of the work
Another 30-40% is jamming and pairing directly with engineers
A real slice is now actual implementation, building and shipping

“as a designer a few years ago I would say like maybe 60 to 70% of it was like mocking and prototyping stuff up”
Jenny Wen · 17:30

“now I feel like the the mocking up part of it is like 30 to 40% and then there's that other 30 to 40% there…”
Jenny Wen · 18:00

#design#workflow#time-management

Takeaway39:30

Design Managers Should Rotate Back Into IC Work

Jenny went back to full-time IC work at Anthropic and says it taught her hard skills she'd never have picked up while only managing, plus the empathy to guide teams through a rapidly changing process. She argues design should copy engineering, where new managers take a months-long rotation doing real tasks before managing full-time.

Going back to IC gave her hard skills she couldn't have gained while only managing
It's hard to empathize with how teams work now if you aren't testing the tools yourself
Engineering often makes new EMs and directors take a months-long rotation first
Design has been too people-management oriented and should adopt a similar rotation

“they actually make the EM like take a rotation for a few months and and pick up a few tasks and really understand how the…”
Jenny Wen · 40:00

#management#career#ic#leadership

Takeaway57:30

Why Teams That Roast Each Other Perform Better

Jenny doesn't force teams to roast each other, but sees it as a good sign of psychological safety, the same comfort you have poking fun at friends. The balance is pairing that safety with clearly high standards, which she likens to being a tough parent and to radical candor: care deeply and challenge directly.

Roasting isn't mandated but signals people feel safe and comfortable with each other
It's also healthy when reports feel safe poking fun at you as a leader without fear
Balance psychological safety with clearly high standards, which she compares to a tough parent
Safety makes applying high standards easier because you can do it without fear
Lenny ties it to radical candor: caring deeply and challenging directly

“it can be a really good sign when uh the people on your team kind of feel comfortable just like kind of poking fun at…”
Jenny Wen · 58:30

“you also have to make sure that you they know that you have really high standards.”
Jenny Wen · 1:00:00

#team-building#psychological-safety#management#culture

Key ideas

Frameworks in this episode

AI Introspection Rubric

Building Trust Through Speed

Deliberate Low-Leverage Leadership

Legibility Framework for Spotting Frontier Ideas

Roasting Plus High Standards

Stratified Design Work

Three Designer Hiring Archetypes

People & resources mentioned

People · 11

Resources · 29

Insights & moments

⚡Myth Buster· 2

Why Figma Still Matters in a Code-First World

Co-work Wasn't Really Built in 10 Days

◆Hot Take· 3

The Classic Design Process Is Basically Dead

AI Will Get Better at Taste, But Someone Must Be Accountable

Low-Leverage Time Is Actually High Leverage

✶Explainer· 4

Design Work Has Stratified Into Two Types

You Can't Mock Up Non-Deterministic AI

Shipping Early Without Losing Trust

The Three Designer Archetypes Jenny Hires For

❝Story· 1

The Best AI News Is Inside the Lab's Slack

?Q&A· 1

Would You Hire Claude as a Designer?

✦Tool· 1

Jenny's AI Stack: A Fully Claude House

▲Takeaway· 3

How a Designer's Time Split Has Changed

Design Managers Should Rotate Back Into IC Work

Why Teams That Roast Each Other Perform Better

Myth Buster· 2

Hot Take· 3

Explainer· 4

Story· 1

Q&A· 1

Tool· 1

Takeaway· 3