Meta scales back staff cuts and AI agents as error cleanup eats into employee time

Reuters reported that Meta scaled back plans to axe staff and use more AI agents ^[11]. The rebuild of its managerial ranks is a separate account, drawn from LinkedIn posts by workers ^[12]. The thinning it is now reversing dated from a 2023 efficiency push ^[12], and the Guardian places the two developments side by side without stating that one caused the other. Meta did not respond to requests for comment, and Microsoft and Walmart declined to comment ^[2].

Microsoft's own account is the most detailed in the record, and it puts the problem in process. Kathleen Hogan, its chief strategy and transformation officer, said in a recent blogpost: "We initially treated AI like a traditional technology rollout: deploy the tools, provide training, drive adoption." She also said: "We learned that access and usage do not equal transformation." ^[5] Hogan's blogpost reads that Microsoft found "adding agents to a broken process still leaves a broken process" and "speeding up one step just creates a longer queue at the next" ^[6]. The company also found that training, particularly pairing early-career engineers with more seasoned employees, boosted productivity ^[6].

The record also shows a skills gap. An Indeed survey published in July found that more than half of employees said they "aren't getting the AI training they need", and just 45% of managers said they felt "equipped to effectively lead AI natives" ^[7]. Johnny C Taylor Jr, president and CEO of the Society for Human Resource Management, said C-suite leaders have a sense that "we're driving the truck and building it at the same time, and that's a problem" ^[8]. Companies are moving from one-off webinars toward "sustained, role-specific training", Taylor said ^[10].

The cost of the correction has not been priced. The reporting does not include a dollar figure for review time, retraining or rehiring. The two surveys measure other things: incidents in the EY survey ^[3] and productivity gains in the Atlanta Fed paper ^[4]. They point the same direction as the company stories.

In our view, the line to budget is review and management time spent on AI output, and it belongs in the forecast before a headcount saving is booked. A handful of company accounts do not establish a pattern on their own, and we would hold that line loosely until a company confirms its own numbers. The sequence in the Meta report is still instructive: a plan to cut, a finding that staff were spending more and more time on cleanup, and a plan scaled back ^[11].