Meta pauses 'Project OT' AI restructuring as productivity gains fail to materialize
Mark Zuckerberg halted a plan to replace thousands of professional roles with AI agents after internal data showed rising technical failures and employee backlash.
Meta has paused or abandoned a sweeping restructuring plan known as "Project OT" (Organization Transformation), which sought to replace significant portions of its human workforce with AI agents. CEO Mark Zuckerberg halted a second wave of layoffs scheduled for November after expected productivity gains from agentic AI failed to manifest and employees protested the use of invasive surveillance tools.
The initiative aimed to transition specialized roles—including engineers, designers, and product managers—into general-purpose "builder" roles overseen by AI agents. Under this "AI native" structure, Meta envisioned organizing staff into small human "pods," with some teams facing potential headcount reductions of up to 60%. However, internal metrics revealed a stark gap between AI development and actual output: while code changes to AI platforms surged 220% year-over-year, improvements to user-facing features rose by only 36%.
The cost of automation
The push toward automation appeared to degrade operational stability rather than enhance it. According to internal data, technical and security incidents increased by 40%, while the time required to resolve those issues grew by 70%. These figures suggest that the complexity of Meta's corporate environment resisted the simplified automation goals of Project OT.
Cultural friction further derailed the plan. To train the AI agents intended to replace staff, Meta installed software to track the mouse movements and keystrokes of U.S. employees. This invasive monitoring led to a 19-point drop in employee sentiment scores, fueling internal protests against the company's approach to AI training.
Industry implications
The failure of Project OT highlights a critical disconnect between the industry hype surrounding "agentic AI" and its practical application in large-scale professional environments. It demonstrates that even a company with Meta's vast infrastructure and capital cannot simply automate away thousands of high-skill roles when the underlying technology fails to accelerate as predicted.
Zuckerberg acknowledged the shortfall, noting that the trajectory of agentic development over the preceding four months had not accelerated in the way the company expected. The episode serves as a cautionary tale for other tech giants attempting to leverage AI for aggressive workforce reduction.
What remains
While the November layoffs were paused, it remains unclear if Project OT is permanently dead or merely being recalibrated. Meta continues to invest heavily in AI infrastructure, but the company must now navigate a damaged relationship with its workforce and a technical reality where AI-driven code is not yet translating into efficient product delivery.