Artificial-intelligence “agents” are moving from experimental pilots to mainstream corporate tools, with integration extending well beyond customer service chatbots. At Workato’s World of Workato event on 22 August, the automation platform introduced an “Agent Knowledge Graph” that links AI agents to a company-wide data fabric and added support for a Model Context Protocol so agents can speak to each other through application-programming interfaces. Workato says more than 80 per cent of its customers’ agent projects reach full deployment, contrasting with an MIT survey that found only 5 per cent of AI pilots deliver rapid revenue gains. Adoption is being driven from the top down. A new poll by cybersecurity firm CalypsoAI found that 67 per cent of chief executives would use AI even if it violated internal rules, compared with 52 per cent of rank-and-file staff. Yet 38 per cent of CEOs—and 30 per cent of employees—conceded they still do not understand what an AI agent is, underscoring a knowledge gap even as deployment accelerates. The technology’s advance is already reshaping corporate hierarchies. U.S. universities and recruiters report a jump in the number of workers reporting to each manager, while recent mass layoffs at Microsoft and Amazon have disproportionately targeted middle-management roles in what academics call the “Great Flattening.” Babson College professor Peter Cohan warns that AI can now translate high-level strategy into actionable tasks, a core middle-manager function. Some observers argue even the C-suite is not immune. Former Google X executive Mo Gawdat told the Diary of a CEO podcast that smarter systems will eventually outperform many corporate leaders and could run businesses built by just a handful of humans. Others, including Intercom chairman Eoghan McCabe, see opportunity for employees who master the new tools, but agree that reskilling and robust governance will be critical as autonomous agents gain decision-making authority.