Meta reportedly explored one of Silicon Valley’s most aggressive workplace transformations yet, with plans to make the company “AI native” by deploying autonomous AI agents, dramatically shrinking some teams and redesigning how employees work. But the ambitious initiative ran into employee resistance, technical problems and disappointing productivity gains, according to a Reuters investigation published on August 26.
Meta Considered Cutting Some Teams by Up to 60%
The internal initiative, code-named “Project OT” for Organization Transformation, was developed in early 2026 under CEO Mark Zuckerberg.
Internal scenario planning examined reducing the size of some Meta teams by as much as 60%, with AI agents taking over portions of work previously handled by employees. Meta stressed that this did not mean it planned to eliminate 60% of its overall workforce; the figure applied only to some of the most drastic scenarios considered for individual teams.
The proposed model envisioned smaller groups of highly skilled employees working alongside AI. Traditional product teams of roughly 10–20 people could, in some cases, be replaced by compact 3–5-person “AI-native” pods with more fluid roles.
AI Agents Fail to Deliver Expected Productivity
The transformation encountered a fundamental problem: AI technology was not producing the efficiency gains executives had expected.
Internal figures reviewed by Reuters showed that AI helped employees generate far more code, but the increase did not translate proportionally into improvements reaching users. Code changes to internal software and infrastructure rose 220% year over year, while changes resulting in new or upgraded user-facing features increased only 36%.
There were also signs of operational trouble. Internal posts warned that AI agents were carrying out large-scale disruptive actions, while major technical and security incidents reportedly increased 40% and time spent dealing with such problems rose 70%. Meta declined to comment to Reuters on those internal figures.
Employee Morale Drops Amid AI Replacement Fears
The restructuring also generated substantial anxiety among Meta employees, many of whom feared they were effectively being asked to train AI systems that could eventually replace their jobs.
Employee sentiment fell sharply. Meta’s internal Pulse survey reportedly showed favourable employee sentiment declining from 74% to 55% amid layoffs, restructuring and concerns about the company’s AI strategy.
Meta went ahead with a workforce reduction of about 10% in May, but Zuckerberg cancelled planning for a second restructuring wave that had been expected in November.
Zuckerberg Acknowledges AI Progress Was Slower Than Expected
In an internal town hall in July, Zuckerberg acknowledged that the development of AI-agent technology had not accelerated as quickly as he had anticipated, according to Reuters. He nevertheless said he expected the technology to improve and deliver greater benefits over the following months.
Meta has since emphasized a less disruptive message around AI, portraying the technology as a way of empowering people rather than simply automating their jobs.
The company confirmed Project OT existed but said it was a broader restructuring effort involving cost reductions, team redesign and redeployment of workers into priority areas. It also said many of the scenarios examined during planning were never implemented.
Meta Still Betting Heavily on AI
The setback does not mean Meta is abandoning artificial intelligence. The company continues to invest heavily in AI models, infrastructure, agents and products while reorganizing parts of its workforce around the technology.
Meta plans to invest at least $130 billion in AI chips and other infrastructure this year, according to Reuters, increasing pressure on the company to demonstrate tangible returns from its enormous AI spending.
The experience offers an important reality check for the broader technology industry: AI may transform corporate work, but replacing large numbers of employees with autonomous agents remains far more complicated than simply deploying the technology.
About the Author
