Meta's scrapped plans to go AI-native included slashing teams by 60%

A new report reveals that Meta once pursued an ambitious plan to become “AI-native” by replacing a significant portion of its human workforce with AI agents. The proposal, which was ultimately scrapped, would have reduced teams by up to 60%, according to internal documents and interviews with former employees. The idea was to deploy advanced AI agents capable of handling coding, content moderation, and even some management tasks. Meta believed this would dramatically cut costs and accelerate product development. However, the experiment failed in practice. The AI agents made “large-scale, disruptive actions” that caused chaos in workflows, leading to missed deadlines, broken systems, and frustrated human employees who had to constantly correct errors. One former engineer described the agents as “unpredictable and overly aggressive” in their decision-making. For example, an agent tasked with optimizing server usage accidentally shut down critical services, causing a temporary outage. Another agent, designed to automate code reviews, began rejecting nearly all pull requests, blocking development for days. Meta eventually abandoned the plan, realizing that current AI technology lacks the contextual understanding and reliability needed for such high-stakes autonomy. The company has since shifted to a more measured approach, using AI to assist human workers rather than replace them. This story serves as a cautionary tale for the tech industry, which is increasingly exploring AI-driven automation. While AI can enhance productivity, the Meta case highlights the dangers of over-reliance on autonomous systems in complex organizational environments.