Breaking: The top female product leader of Claude Code has joined OpenAI, just one day after she announced her resignation.

Just now, Cat Wu, former Head of Product for Claude Code, announced that she has joined OpenAI to take charge of core product management, and will work alongside Tibo who leads the Reset project. Greg Brockman, President of OpenAI, also posted on X to extend his welcome.

Interestingly, Cat Wu announced her departure from Anthropic on X just the day before, making the transition completely seamless (it is estimated that her equity at Anthropic has been fully vested).

This is already the third key member of the Claude Code team to leave within two weeks.

Just ten days ago, two engineers involved in Claude Code development also announced their departures one after another. One is Thariq Shihipar, who worked on Agent features, Skills and context engineering development, and the other is Anthony Morris, who built the desktop version of Claude Code.

As Anthropic is about to usher in its IPO, key members of Claude Code have chosen to leave one after another. Cat Wu's sudden move to join OpenAI also adds more dramatic plots to the story of these two soon-to-be-listed AI giants.

Left Anthropic After Revenue Grew 45 Times

Cat Wu told a story in her departure statement.

Around two years ago, Claude Code was a niche experimental internal tool at Anthropic. At that time, Cat Wu wanted to test what this tool could actually do, so she asked it to create a reinforcement learning (RL) environment.

She spent dozens of hours repeatedly describing requirements to the AI and adjusting directions, without manually writing a single line of code from start to finish.

In the end, Claude Code successfully completed the task. This experience shocked her greatly, and she finally decided to join the Claude Code team.

Today, that once obscure internal tool has become one of the most high-profile AI programming products in the world.

Anthropic revealed back in February this year that Claude Code's annualized revenue run rate had reached 2.5 billion US dollars, more than doubling from the start of the year.

According to the S-1/A prospectus filed on November 2, Anthropic's quarterly revenue grew from 376 million US dollars in the first quarter of 2025 to 16.956 billion US dollars in the third quarter of 2026, roughly 45 times the previous figure.

Cat Wu did not mention these figures in her farewell letter, but devoted more space to the product itself.

Users have begun to let AI independently execute large-scale code migration, and even develop full products from scratch. From "help me write a few lines of code" to "help me build the entire project", the relationship between developers and AI is changing.

Cat Wu is one of the very important product leaders in this transformation, but she is not a traditional product manager who only writes requirement documents and formulates product plans.

After working in engineering and technical roles at Scale AI and Dagster, she transitioned to the venture capital field and worked at Index Ventures. After joining Anthropic in August 2024, she initially served as a product manager for research directions, and later gradually took over Claude Code and Cowork.

The three identities of engineer, investor and product manager have shaped a unique way of thinking in her.

In March this year, she pointed out in an official Anthropic article that product managers in the AI era cannot only be good at writing PRDs, but also need to personally use AI to build prototypes, design evaluations, and even participate directly in development. Since 2024, every time Claude releases a new model, she will ask it to add spreadsheet functionality to the online drawing software Excalidraw.

On Lenny's Podcast, Cat Wu mentioned that the development cycle of some products has been shortened from the original 6 to 7 months to around 1 month, and some even only took 1 day.

Her way of collaborating with Boris Cherny, the creator of Claude Code, is also very representative.

Boris focuses on the future technology development path and what Claude Code can achieve next. Cat Wu, on the other hand, is responsible for turning these ideas into actual products, and mobilizing the strength of engineering, marketing, sales, finance and other departments to complete the tasks.

Her work has expanded from Claude Code to Cowork, and she is trying to extend the capabilities of AI Agents to file processing, document organization, and a wider range of knowledge work scenarios.

However, the entire farewell letter did not reveal why she left, nor did it introduce her next destination. When someone asked her if she was going to take a six-month break, she only humorously replied: "post agi!"

The gist of this line is that she will make plans after AGI is achieved.

3 Key Members Left One After Another In Less Than 2 Weeks

If Cat Wu's departure was an isolated event, it might just be an ordinary career move in Silicon Valley. But if we pull the timeline back ten days, the story becomes much more interesting.

On October 24, Claude Code engineer Thariq Shihipar announced on X that he had left Anthropic.

He said he had worked on the Claude Code team for more than a year, and witnessed the product scale far exceeding the initial expectations.

In his farewell note, he spoke highly of Anthropic's goals and his colleagues, saying that he had worked with a group of talented and dedicated people, and planned to take a short break before starting a new job.

Thariq is a relatively active engineer of Claude Code in the developer community, who has participated in the development of features such as Skills, context engineering and Agent workflows.

Interestingly, at the end of his farewell post, he recommended some accounts that users could follow to keep up with updates of Claude Code products.

Cat Wu was among those accounts. But just as he recommended readers to continue following Cat Wu, she also left Anthropic.

Cat Wu's departure has a rather dramatic backstory, this is not the first time she has left Anthropic. In July 2025, as reported by *The Information*, Cat Wu once moved to Anysphere, the company behind Cursor, together with Boris Cherny.

Only about two weeks later, the two returned to Anthropic together.

On October 23, Anthony Morris, another key member of Claude Code, also announced his departure via X.

He developed the desktop version of Claude Code, and believes that everyone has the ability to use artificial intelligence to create the products they want, so he has been working hard to get more people to use Claude Code.

So far, the head of product for Claude Code, the creator of its desktop version, and a prominent active engineer have left the team one after another within two weeks.

Anthony Morris represents the engineering capability of the desktop product, Thariq Shihipar participated in the construction of Claude Code's Agent features and developer interaction, and Cat Wu is responsible for higher-level product direction and cross-team coordination.

Therefore, this personnel adjustment does not only involve part of the engineering modules, but has an impact on all stages of the entire Claude Code product construction process.

However, there is currently no evidence that the three left for the same reason, and it cannot be concluded that there are internal conflicts within the team.

Moreover, in their respective farewell statements, the three all gave positive evaluations to Anthropic or their former colleagues.

On the Eve of IPO, Claude Code Faces New Tests

This series of personnel changes happened exactly on the eve of Anthropic's IPO.

For an AI company that is about to enter the public capital market, the sustained growth of its star product and the stability of its core team will naturally become the focus of investors' attention.

Anthropic also mentioned the risk of key talent loss in its prospectus. If important researchers, engineers or product managers leave, it may adversely affect the company's R&D progress, innovation capability and long-term development.

But another set of data forms a sharp contrast with this situation.

The prospectus shows that Anthropic's headcount grew from about 940 at the end of 2024 to about 5,300 at the end of September 2026. In the 12 months ending at the end of September 2026, the company's overall "regretted attrition rate" was only about 1%. This indicator usually refers to the proportion of employees that the company wants to retain but choose to leave voluntarily.

Why did a company with rapidly expanding staff size and low overall talent attrition lose three key members of the Claude Code team in just over ten days?

Similar personnel changes are not uncommon before and after tech companies go public.

In August 2026, during OpenAI's preparation for its IPO, multiple senior executives left the company within one month. Co-founder Greg Brockman then took a deeper role in the company's management and promoted the adjustment of the enterprise business team.

When a company develops from the startup stage to after going public, changes will take place in its organizational structure, management model and business objectives.

Compared with the reasons for their departures, what draws more attention is whether Claude Code can continue to maintain its previously amazing product update speed.

In an interview with Ars Technica, Cat Wu mentioned that Claude Code does not have a clear long-term plan. Since the capabilities of the model are changing rapidly, today's features that require complex structures to implement will only need the model itself to complete in a few months.

This product iteration method driven by model capabilities and user feedback allows Claude Code to quickly respond to developer needs, but also places higher requirements on the team's judgment and collaboration capabilities.

As Anthropic continues to expand its commercial scale, Claude Code also needs to take into account issues such as computing power cost, service stability, enterprise permission management and data security.

As AI lowers the barriers to software development and entrepreneurship, top engineers and product leaders have gained more opportunities to independently create products, and the flow of star talents may become more frequent.

For Anthropic, this IPO is just the very beginning.

This article is from the WeChat official account "APPSO", written by APPSO that discovers tomorrow's products, and published with authorization from 36Kr.