Meta's AI Ambitions
In a bold move, Meta attempted to transform into an "AI native company" by replacing a significant portion of its workforce with AI agents. The initiative, codenamed Project OT, aimed to create "small teams of high performers with AI agents doing the work underneath them." > The idea was to make meta what they called an AI native company. — Trent Cotton Despite initial layoffs in May affecting 10% of the workforce, the second phase was abruptly halted by Mark Zuckerberg, who cited that "the technology had not accelerated the way that he thought that it would." > He later said that the technology had not accelerated the way that he thought that it would. — Trent Cotton
What Went Wrong?
The episode delves into the metrics that indicated the project's challenges. Code changes at Meta increased by 226% year-over-year, yet user-visible features only rose by 36%. > Code changes at meta went up 226% year over year. Features that users could actually see went up 36%. — Trent Cotton Security incidents also spiked by 40%, and internal sentiment plummeted from 74% favorable to 55%. > Security incidents went up 40%. Internal sentiment dropped from 74% favorable to 55%. — Trent Cotton These numbers suggest that while technical output increased, the quality and security of the work did not improve proportionately.
The Role of HR Mindsets
Host Trent Cotton, with over 20 years of HR experience, discusses the importance of different mindsets in addressing such organizational challenges. He emphasizes the need for an "architect mindset," which focuses on understanding the foundational structure of the company rather than just cutting headcount. > If you want an AI native company, you're not necessarily always cutting staff. You are rebuilding how work gets defined, assigned, reviewed, and shipped. — Trent Cotton Cotton also highlights the "engineer mindset," which is concerned with the handoffs within an organization, and the "scientist mindset," which questions whether metrics are truly predicting business outcomes.
Lessons for the Future
Cotton argues that the failure of Project OT underscores the necessity for HR to be involved in strategic decisions. He suggests utilizing tools like the "value of work matrix" to better align human capital with organizational goals. Ultimately, the episode serves as a cautionary tale about the pitfalls of over-reliance on AI without thorough structural and strategic planning.
