Meta and Block are replacing traditional middle managers with flatter structures and AI-assisted roles, signaling a fundamental shift in how tech companies organize work.
Jack Dorsey wants everyone at Block to report to him eventually. Mark Zuckerberg says a single talented engineer, armed with AI, can now do the work of an entire team. Two of the most influential figures in tech are not just trimming managerial layers: they are attempting to rewire what a company actually is.
Meta's Reality Labs unit has already started replacing conventional manager titles with new labels like "AI builder," "pod lead," and "org lead" as part of a push toward becoming what the company calls "AI-native." The idea is that routine managerial tasks, from performance reviews to coordination, can be handled by AI systems, freeing human leads to focus on strategic direction. Meanwhile, Block is stripping the word "manager" from its vocabulary entirely. Those roles are becoming "player-coaches" who build alongside their teams rather than overseeing them from above, while individual contributors are expected to lean on AI tools to make decisions independently.
As Business Insider reported, job postings for middle-management roles dropped 12.3 percent in 2025 compared to the previous year, according to data from Indeed. That decline outpaces an overall softening in job listings, suggesting that the managerial layer is being deliberately targeted.
The war on middle management did not start with AI. Companies have spent years flattening org charts in pursuit of speed. The logic is straightforward: every layer between a decision-maker and the person executing adds friction, delays feedback, and dilutes accountability. Google, Amazon, and others have periodically culled managerial ranks over the past decade, arguing that leaner teams move faster.
What makes this moment different is that AI gives companies a credible replacement for the coordination and oversight work that middle managers traditionally handled. Scheduling, status tracking, initial performance assessments, and resource allocation are all tasks that large language models and agentic AI systems are increasingly capable of managing. Zuckerberg made this explicit on Meta's January earnings call, noting that the company is investing in AI-native tools specifically to flatten its structure.
Dorsey is more aggressive. In a blog post co-authored with Sequoia partner Roelof Botha, he wrote that "there is no need for a permanent middle management layer." On a Sequoia podcast, he said his goal is to reduce the layers between himself and Block's roughly 6,000 employees from about five to two or three this year, with the ultimate aspiration of having no layers at all.
What History Teaches About Flat Organizations
This is not uncharted territory, and the map has warning signs. Zappos famously adopted holacracy in the 2010s, eliminating traditional managers and job titles in favor of self-organizing circles. The experiment was chaotic. Turnover spiked, key executives left, and the company struggled to maintain strategic coherence. Zappos eventually walked back many elements of the system.
Chris Kaufman, a cofounder of StockX and leadership consultant, is skeptical that rebranding managers solves much. He compares it to "moving the peas to the corn compartment in a TV dinner," arguing that without real structural change, new titles are cosmetic.
Harvard Business School professor Linda Hill, who cofounded the startup InnovationForce, offers a more measured view. She says small, cross-functional teams can accelerate innovation, but they still need what she calls "bridgers," people who facilitate collaboration across groups with different expertise. Remove that connective tissue entirely, and you risk creating efficient silos that cannot align on shared goals.
The Practical Reality for Startups and Founders
For early-stage companies watching these experiments unfold, the takeaway is not to fire every manager tomorrow. It is to audit what your managers actually spend time doing. If the answer is mostly coordination, scheduling, and report generation, AI can likely take on a significant portion of that workload today. If they are coaching talent, resolving interpersonal conflicts, and making judgment calls about trade-offs, those responsibilities remain stubbornly human.
The companies that get this right will not simply eliminate layers. They will redesign roles around what humans do best and let machines handle the rest. The ones that get it wrong will find that flat organizations without strong connective leadership do not move faster. They just fracture more quietly.
