AI Pulse · AI Trends Pulse

Meta reportedly planned a year-long restructuring called “Project OT”, built around the idea that AI agents could replace a large share of work across some teams. The plan was meant to happen in two layoff waves, first in May and then in November. The May wave went ahead and cut 10% of staff. But the November wave was cancelled after employee backlash, weak internal evidence of productivity gains, and a rise in critical crashes linked to the AI-led changes, according to this report.

For business owners, the useful lesson isn’t that AI agents don’t work. It’s that replacing people at scale is a much higher bar than automating a task. An agent might draft a report, route a customer request, or pull data from several systems. But if it creates failures that need constant checking, rework, or emergency fixes, the apparent labour saving disappears quickly.

Treat AI as a controlled operating change, not a headcount target. Start with a narrow workflow, measure time saved and error rates, keep a clear human owner, and only expand once the process holds up under real volume. This is the kind of thing we build into an AI command centre, where teams can see what automation is doing and intervene before small failures become expensive ones.