Mark Zuckerberg’s reported effort to reshape Meta around artificial intelligence offers a revealing test of a question confronting companies far beyond Silicon Valley: can AI meaningfully reduce the need for human staff without weakening the work those employees perform? Meta’s experience suggests that, while AI may change how organisations operate, replacing large segments of the workforce remains far more difficult than presenting an ambitious automation strategy.
The Vision Behind Project OT
Reuters reported that Meta explored an internal restructuring initiative known as Project OT, short for Organization Transformation. The proposal envisaged an “AI native” organisation in which AI systems could assume much of the routine work undertaken by thousands of employees, while smaller groups of highly skilled staff supervised those systems. Internal planning exercises considered reducing the headcount of some teams by as much as 60 percent, although Meta said this figure related to scenarios under review rather than a company wide target.
The project reportedly aimed to redefine work across technical and product functions. Rather than operating through large, specialised teams, Meta considered moving towards smaller teams of broadly capable employees supported by AI agents. This approach was intended to speed up execution, reduce layers of management, and concentrate resources on priority AI projects. In theory, such a model could allow the company to maintain output with fewer employees while allocating more investment to advanced computing infrastructure and AI development.
Why the Strategy Became Difficult
The plan encountered practical and organisational constraints. Reuters reported that Meta proceeded with a global workforce reduction of about 10 percent in May and reassigned around 7,000 employees to initiatives connected to AI workflows. However, the more expansive restructuring plans were scaled back, including a proposed later wave of changes.
The central difficulty was not simply whether AI could generate code, draft analyses, or complete repetitive tasks. It was whether those outputs could be reliably integrated into complex organisational processes. In large technology companies, work often depends on institutional knowledge, coordination across teams, risk assessment, product judgement, and accountability when systems fail. AI tools can support these tasks, but replacing the people responsible for them creates new demands for quality control and oversight.
Zuckerberg reportedly acknowledged in an internal meeting that progress in AI agents had not accelerated as quickly as anticipated. This is significant because workforce reductions based on projected automation gains can be difficult to reverse if the underlying technology does not deliver the expected productivity improvements.
Implications for Corporate AI Adoption
Meta’s case highlights a broader distinction between AI augmentation and AI substitution. Augmentation means employees use AI to complete work more efficiently, potentially allowing firms to redesign tasks and improve productivity. Substitution, by contrast, assumes that AI can take over work sufficiently well that organisations can remove human roles at scale. The latter requires a much higher level of reliability, governance, and operational maturity.
For companies considering similar transitions, the episode underscores the importance of testing AI systems in narrowly defined settings before connecting staffing decisions to uncertain productivity forecasts. Automation can produce efficiencies, but it can also create hidden costs through errors, duplicated review processes, security concerns, employee disruption, and reduced organisational resilience.
The response also matters internally. Major workforce changes can affect morale, trust, and the willingness of employees to share expertise during periods of transition. Where a company’s strategy depends on skilled workers collaborating with AI systems, retaining confidence among those workers may be as important as the technical capabilities of the systems themselves.
A Cautious Lesson for AI Native Workplaces
Meta’s Project OT does not mean AI driven organisational change has failed. Instead, it illustrates that the path toward AI intensive workplaces is likely to be gradual, uneven, and dependent on the nature of each task. AI may increasingly reshape roles and workflows, but Meta’s reported retreat from its most ambitious scenarios shows that human judgement remains central when technology is deployed at organisational scale.
