Meta’s Project OT imagined an 'AI-native' company where AI agents would take over much of the work done by thousands of employees, according to a Reuters investigation cited in the source. Teams could become dramatically smaller, some management layers could disappear, and Zuckerberg has been using what Meta calls a 'CEO agent' to retrieve answers that previously required going through several layers of staff.
The appeal is easy to see. Meetings take time, information gets lost between layers, and decisions move slowly. AI can compress all of that. In the words of the source, AI can reduce 'coordination friction' by handling scheduling, documentation, data reconciliation, approvals and communication.
But the same source draws a harder line around 'cognitive friction' — the resistance between ideas that comes from disagreement, different assumptions and inconvenient facts. That friction can be uncomfortable. It can also be useful.
The risk is that an AI agent becomes a sophisticated mirror. Instead of forcing distance from an existing worldview, it can make that worldview more articulate, comprehensive and persuasive. The higher someone rises in an organization, the more information is filtered before reaching them. An AI system can speed up that flow while also reducing exposure to the people who would have challenged the assumptions behind it.
The source argues that AI should function more like a sparring partner than a mirror. It uses the language of 'peacemaking' versus 'sensemaking': one smooths disagreements into a clean answer, while the other exposes contradictions, hidden assumptions and evidence that does not fit. For a CEO, the second role may be the more valuable one.
Meta ultimately abandoned its most aggressive workforce-reduction plans after internal data raised questions about whether dramatically increased AI-assisted coding was translating into comparable gains in user-facing products. That detail matters. Faster output is not the same thing as better output, and the market has already seen enough software promises to know the difference between automation and productivity.
For business leaders, the lesson is bigger than one company. Capital rewards clear rules, but organizations also need dissent, challenge and judgment. If AI becomes a CEO's best yes-person, the company may get leaner while becoming less resilient. The taxpayer is not involved here, but the principle is the same: efficiency alone is not always value creation.
