Charles Rollet takes a deep dive into how tech companies are reorganizing around AI. In particular, Meta relies on AI coding tools to ship more products faster.
There are signs that it’s working. Meta ranks fairly high in revenue per employee, a metric that Silicon Valley is increasingly focused on.
Below is our chat with Charles. Edited for length and clarity.
You and other colleagues on the tech team have been writing a lot lately about how generative AI and AI coding tools are changing the way tech companies operate. This is especially true within the meta. What are the big takeaways from this study?
You can no longer hide from AI. Until recently, employees could just be good programmers or product managers. But now it seems impossible to advance in the meta without leveraging these tools and showcasing your AI talents. These requirements have already been formalized, and the company has set clear goals for some of its engineers to generate 50% to 80% of their code with AI assistance.
Why is Mark Zuckerberg promoting this so much at Meta? What are the company’s goals here?
The answer is simple. It’s productivity. Instead of a bunch of young developers running the foundation, Zack wants 100x more engineers to command an army of AI agents. Considering recent advances, it may be possible. However, this also leads to a worrying trend for Meta employees. If AI is so powerful, why does the company need to hire over 76,000 very expensive employees?
Employees who participate in internal AI Weeks may be digging their own grave, former director of metaengineering Eric Meyer recently told me. He worries that a company with as many users as Meta won’t be able to launch more than 10 times as many new features. Their user base just can’t absorb it. Therefore, the logical alternative would be to reduce headcount.
What parts of Meta’s business are changing the most due to these “AI at Work” changes?
Reality Labs has the most aggressive AI-based reorganization in the meta that we know of to date. The division’s internal tools team of about 1,000 people has been radically reorganized, eliminating old roles and focusing its work on small AI-native “pods.” The team will rebrand its employees as “AI builders” and managers will become known as “AI pod leads” who use AI to help review performance.
News of the reorganization reached other parts of Meta, with some fearing that the “AI pod” model would be more widely adopted and used as a justification for layoffs. Mr. Mehta said in his reorganization memo that employee numbers would not be affected, but this speaks to the concerns of some employees within the company.
Does this work or do you think there is a problem?
You’ll probably run into a lot of issues, like the AI agent accidentally deleting your inbox, but most of them are manageable. Claude Code seems to represent a real step change in AI capabilities, and to Meta’s credit, it doesn’t preclude employees from experimenting with all sorts of tools. Right now, I’m more worried about the atmosphere of startups coding serious problems into their products than I am about well-run companies like Meta.
Is this primarily about efficiency and cost reduction, or is there something else going on?
There are definitely other factors. The big push appears to be encouraging employees to be more flexible. For example, as part of the AI pod reorganization, Mehta told staff he expected engineers to do design work when needed to get things done. This could lead to some interesting products coming out of Meta if roles aren’t too tightly defined and people can build whatever they want. However, I think the important thing is efficiency and productivity. I won’t risk bragging about building an AI agent to do all my work.
Sign up for BI’s Tech Memo newsletter here. Please contact us by email. abarr@businessinsider.com.