Former Claude Design Lead Jenny Wen Joins Musk's Camp, to Lead Grok Bot Design

Claude's former head of design has turned around and joined Elon Musk's camp.

She is Jenny Wen.

Earlier this year, she was still at Anthropic, in charge of design for Claude and Cowork.

On October 23, Musk's Grok Bot launched the Vault.

She reposted this update on X, writing:

This is what we shipped today!

From Claude to Grok Bot, Jenny Wen's turn began with a tweet in July of this year.

That day, she announced three big things at once:

Leaving Anthropic, giving birth to her precious daughter, and about to join Cursor as head of design.

A month later, on August 14, SpaceX officially completed its acquisition of Cursor, and she entered Musk's camp along with the team.

"A misguided engineer"

Who put joy into the whiteboard

In her self-introduction on X, Jenny Wen used three words:

generalist, realist, escapist.

These three words span a wide range, and are even somewhat contradictory, yet placed in the context of her experience, they all make sense.

By generalist, she means her self-proclaimed "weird block-shaped designer," with several skills deep enough to count.

By realist, she means her catchphrase "It is what it is," which sounds bleak but is actually a way of loosening her own restraints.

As for escapist, after finishing a summer's work in New York, she hit the gas and drove across Alaska.

This readiness to turn the page was already there when she was a student.

While studying at the University of Waterloo, she first majored in urban planning, and halfway through, she simply tore it all down and started over, switching to systems design engineering.

She later explained that she found it boring as she went along and wanted a direction that was more challenging and more oriented toward math and science.

During an internship, she worked as a software engineer alongside UX designers, and only then learned for the first time that there was a profession dedicated to figuring out how to make things pleasant to use.

Later, doing design at Dropbox, she joked that she often felt like "a misguided engineer," always trying to convince others, including herself, that she really was a designer.

In 2020, when the whiteboard tool FigJam started from scratch, she was the earliest designer on board.

With no lengthy PowerPoint arguments, she and engineers directly hand-built prototypes of stamps, emoji and cursor chat with real code, and shamelessly barged into colleagues' meetings to ask everyone to try them on the spot.

There were plenty of flaws, but the whole meeting room was laughing.

"We just knew there was something good here." Colleagues laughing out loud at the rough prototype was more convincing than any number of alignment meetings.

In 2023, she served as host on the main stage of Figma's annual user conference Config, facing an audience of about 8,000.

Later, she rose to design director at Figma, leading the design team for products including the FigJam online whiteboard and the Figma Slides presentation tool, with as many as a dozen designers under her.

Stepping down from the 8,000-person stage

Back to the critique table

Having reached design director at Figma, continuing up the management ladder seemed the natural choice.

Yet in January 2025 she left Figma and joined Anthropic as an individual contributor (IC), going back to doing design hands-on.

By then she was already quite clear-eyed: the middle-management rice bowl might not stay steady in the future. Without trying it herself, it would be hard to understand what the team was going through.

Rather than watching the waves from shore, better to jump into the water first.

After returning to the front line, her most unfamiliar lesson was sitting back down at the critique table to take criticism.

Being criticized, and criticized so frequently, is really hard.

She admits that laying your work out for others to pick apart is itself a rather vulnerable thing.

In September 2025, after more than half a year of grinding on the front line, she fired at the traditional design world at the Hatch conference in Berlin, putting forward the thesis that shook Silicon Valley: "The traditional design process is dead."

At the start, she earnestly acted out the textbook process, then joked: "Congratulations, you've solved all the company's problems!"

Then she pivoted and asked: "That's not realistic at all, right?"

She made a biting analogy: the old process is like IKEA instructions—follow them and you can assemble a Billy bookcase; but when you're doing design, you have no idea whether what you end up with is a bookcase, a lamp, or a hot dog.

What's more, she said Anthropic's engineers routinely have "7 Claudes" running at once writing code, and even the engineers are worrying about how to keep up with themselves. If designers then cling to the old process and slowly draw wireframes, they'll only hold things back.

Ironically, this talk mocking the process, just three or four months later, looked outdated to her when she revisited it—"a bit embarrassing."

Later, she cut the time she spent drawing mockups and making prototypes from 60-70% to 30-40%, spending the rest of her time beside engineers or writing code herself.

When someone said in Slack that an icon was crooked, she directly @Claude to fix it and casually merged the PR.

The girl who once joked she was "a misguided engineer" had come full circle and truly pushed herself into being half an engineer.

In the prototype she "really didn't get"

Lies the shadow of Cowork

In January 2026, Claude Cowork burst onto the scene.

The story outsiders love to tell most is the Silicon Valley myth of a small team shipping a hit in a 10-day blitz.

But on the podcast, she personally punctured that filter:

It was absolutely not just 10 days, and far more people were involved.

What Cowork should look like—she had already sketched out that direction back in 2025.

The story begins with an internal Anthropic prototype in 2025 called Claude Studio: a dense interface packed with skills, knowledge, what Claude was working on, and output previews.

Her first reaction was honest: "I didn't know what this was doing. I really didn't get it."

Strangely, researchers and quite a few colleagues were gathered around it, buzzing with enthusiasm.

As someone who also does angel investing, she borrowed the VC world's sense of "readability": ideas everyone gets at a glance are usually not novel enough; the ones you don't understand but that make insiders excited are worth digging into to figure out what's actually good about them.

Later, the prototype team pulled the Skills framework out of that tangled mess.

In Jenny's view, Claude is very good at turning the scattered materials users casually toss over into useful results.

But once a task runs long, users want to know: how does it plan to do this? Which files is it looking at? What step is it on now?

So when designing Cowork, she focused on surfacing two kinds of information: Claude's plan and to-dos, and the context and files it is using.

Putting the process in front of users is what makes them more comfortable handing over the work.

By March, this collaboration extended to phones.

Claude Dispatch, which Jenny helped launch, lets users keep assigning tasks to Claude after leaving their computers.

For example, while out buying coffee, users can send tasks to Dispatch like a text message. Claude Cowork or Claude Code on the computer handles the work, and users can check the results on their phones.

But when asked what she is proudest of about Cowork, the first thing that comes to mind is that the product finally shipped: "We actually shipped it."

She explained that when a product is in development too long, you can only see its flaws. Only by putting it in users' hands do you get the chance to know what it can actually help with, and what to improve next.

Pushing open the "one-way door"

Next stop: SpaceX

Few people know that during those months working on Cowork, she was pregnant with her first child.

On March 23, 2026, she posted a somewhat anti-chicken-soup essay on her personal website: "Mourning This Life."

Before that, she was a grind-setter proud of running at high intensity nonstop; after getting pregnant, her body was the first to betray her ambition.

She was exhausted, sweating more easily than usual on subway platforms; she stopped exercising, and the biceps she had worked hard to build seemed to vanish, and she didn't even want to attend the pottery class she had signed up for.

Hormonal changes kept her from sleeping well at night, and in those sleepless late hours she fell into a deeper fear: worried that having a child had walked her through a "one-way door" she could never return from.

After becoming a mother, would the creativity, curiosity, drive, and ambition she cherished disappear along with it? She even felt that a pregnant woman having such thoughts was almost "unholy."

Fortunately, the streak of realism in her personality helped her reconcile this contradictory and complicated feeling.

She realized: mourning the old days is real, and being moved by the beauty happening inside her body is also real. "Being able to hold two seemingly opposing emotions at once is the most human thing."

Less than four months later, both her life and career had new developments.

From switching majors out of "boredom," to giving up the Figma director title and returning to the front lines at Anthropic, to joining Cursor after becoming a mother, Jenny Wen has never followed Silicon Valley's fixed promotion playbook.

When engineers can already have multiple Claude agents writing code and building product prototypes simultaneously, clinging to old processes and a director title will only make you less valuable the longer you cling.

What can protect your job is the courage to start over after seeing the changes clearly, and the eye to spot opportunities and judge direction amid a flood of information.

References:

This article comes from the WeChat public account "Xinzhiyuan," author: ASI Revelations, editors: Yuanyu, Moses, published by 36Kr with authorization.