Amazon has rolled out an internal tool called MeshClaw that lets employees assign routine work to AI agents across the firm. By logging each delegation, staff can earn points on a company‑wide AI leaderboard, turning mundane tasks into a gamified metric. Critics say the system incentivises quantity over genuine productivity, prompting workers to inflate usage scores rather than focus on value‑adding activities.

The initiative emerged as Amazon pushes a broader AI‑first strategy, aiming to embed large‑language models across its logistics, retail and cloud divisions. Management touts MeshClaw as a way to surface inefficiencies, yet internal surveys reveal many teams treat the leaderboard as a performance‑gaming exercise in early 2024. Roughly a third of participants admit they route trivial requests solely to boost their rankings.

Investors watch the experiment because gamified AI usage could distort cost metrics that feed Amazon’s quarterly reporting. If inflated activity masks true automation savings, analysts may overestimate margin improvements tied to the company’s AI investments. For now, senior leaders have pledged to refine MeshClaw’s scoring algorithm, but the episode underscores the tension between internal incentives and transparent performance measurement in the short term.