Tech Giants Hire AI to Manage Other AI Because Humans Clearly Failed

KEY POINTS

•Anthropic researchers Barry Zhang and Mahesh Murag advocated 'agent skills' at December's AI Engineering Code Summit to improve AI efficacy.
•Since launch five weeks ago, thousands of skills have been created by professionals from various non-tech domains to boost agent expertise.
•Industry leaders like Sam Altman and Asha Sharma discussed AI agents replacing junior jobs and flattening organizational hierarchies in 2025.

Amid a race fueled by startups inflating chatbot hype since at least May 2025, Anthropic researchers Barry Zhang and Mahesh Murag swooped in at the AI Engineering Code Summit last month with a revolutionary 'agent skills' plan: instead of flooding the market with flood after flood of AI agents, stack one general agent with 'folders' of knowledge. These skill-folders, created by over a thousand diversified pros (accountants, legal eagles, recruiters—because tech alone can't save us), serve as playbooks teaching AI agents the crippling expertise they still lack, helping companies like Fortune 100s simplify their sprawling, many-layered org charts, much to Microsoft product lead Asha Sharma's delight. Meanwhile, OpenAI CEO Sam Altman touted AI agents replacing junior employees by June; investors at a16z criticized scammy startups upselling chatbot wrappers as AI agents since May; and all this despite today’s shiny agents often failing context tests more embarrassing than your last Zoom call.