Meta Scraps AI-Usage Performance Metric While Pushing Agentic Tool Hatch to Employees
The AMW Read
Meta reverses an internal AI-usage performance mandate amid litigation while piloting its new computer-use agent Hatch, an incremental workplace-policy and product update for an already-known player rather than a structural shift.
Meta Scraps AI-Usage Performance Metric While Pushing Agentic Tool Hatch to Employees
Meta told employees this week that performance reviews will no longer be tied to how much they use AI tools, ending an "AI-driven impact" policy that had rewarded workers with labels like "AI Native," "AI First," and "AI Enabled" based on chatbot and agent usage. The reversal follows a lawsuit filed in July by roughly two dozen former employees alleging the metric contributed to discriminatory May layoffs, including against staff on health or family leave who could not accumulate usage credits. Engineers were told the company will not use AI adoption dashboards or token counts to judge impact, per The Information. Simultaneously, Meta is encouraging (not requiring) staff to test Hatch, an agentic tool that can autonomously browse the web and operate other applications, which employees compare to the viral OpenClaw. Hatch has been running on corporate devices for several weeks ahead of an expected public release, though some employees are wary of connecting it to personal accounts given lingering distrust from a since-paused Meta project that logged employee keystrokes for AI training data.
The shift shows a major AI lab pulling back on usage-based mandates precisely as it introduces its most ambitious internal agent yet, evidence that adoption-metric pressure created legal exposure and morale costs that outweighed its value as a usage-forcing function. Notably, Meta workers say Hatch already consumes more tokens than standard chatbots and coding assistants, so removing the mandate may not meaningfully reduce compute draw as the tool spreads organically. The episode lands alongside Meta shelving a planned layoff round beyond the 8,000 cut in May, reportedly after AI-linked software bugs and slower agent-development progress gave leadership pause.
For builders and investors, Hatch signals Meta fielding a computer-use agent internally before any public launch, competing in the same autonomous browser/desktop-automation category as OpenClaw; teams designing internal AI rollouts should note Meta's lesson that usage quotas invite legal and trust risk independent of a tool's actual utility, and that agent-style workloads carry materially higher token costs than prior chatbot-era baselines when planning infrastructure spend.
