On March 31, 2026, Jack Dorsey published an essay titled “From Hierarchy to Intelligence.” The argument was blunt: corporate hierarchy has always existed to route information through organizations too large for any single person to oversee. Managers aggregate context from below, relay strategy from above and maintain alignment across teams. Artificial intelligence (AI) can now perform those functions continuously and at scale. The messenger, they argued, is no longer necessary.

Unbeknownst to many, the organization chart structure modern American uses is actually quite archaic. Hence, Dorsey wanted to revamp this 2,000-year old tradition. Block cut approximately 4,000 employees and restructured around just three roles: individual contributors who build the system, directly responsible individuals (DRIs) who own specific outcomes on 90-day cycles and player-coaches who remain hands-on while developing people. No permanent management layer.

Whether or not every company follows Block’s playbook, the direction is unmistakable. The organizational middle, the layer of managers, analysts and coordinators whose value was rooted in processing and relaying information, is compressing. For learning and development (L&D) professionals, this shift demands a fundamental rethink of how talent is developed, deployed and retained.

Bees and Beekeepers: A Framework for the Flattened Organization

One useful way to understand this transition is through the lens of bees and beekeepers, a framework from the book, “How to Do More With Less.” In this framework, AI systems are the bees: tireless workers who execute tasks at speed and scale, from data analysis to content generation to workflow coordination. Humans are the beekeepers: the ones who set direction, exercise judgment in ambiguous situations and decide what the bees should be doing in the first place.

The distinction matters because as AI takes over more of the “bee work” that middle managers used to perform (compiling reports, synthesizing updates, routing approvals), those managers face a choice. They can evolve into beekeepers who direct AI systems, provide contextual judgment and focus on work that requires human taste and agency. Or they can resist the shift and find their roles automated out from under them.

This is where the flattening gets personal. When parts of a role get replaced by AI, the person in that role must learn how to redeploy the reclaimed time toward higher-value work. That is not a one-time adjustment. It is a continuous process of self-reinvention, and it requires a capability that most corporate training programs do not develop: the ability and willingness to keep learning.

The Biggest Danger Is Not AI. It Is Apathy Toward Learning.

Walk into any Fortune 500 company running an AI transformation and the pattern is remarkably consistent. Leadership announces a new initiative. Training is rolled out. And the vast majority of employees do the bare minimum to check the box, then return to their existing workflows unchanged.

Across advisory engagements with organizations of every size, the same observation surfaces repeatedly: the employees who thrive in AI-augmented environments are not the ones with the strongest technical skills. They are the ones who are genuinely curious, who enjoy figuring things out, who treat every new tool as an invitation rather than an obligation. They are a small minority. The rest are waiting to be told exactly what to do, completing the minimum required training and returning to their routines. That gap between the curious and the compliant is, in practice, the single greatest risk factor for any organization navigating AI-driven restructuring.

When roles compress and employees need to be redeployed, the ones who can learn a new function quickly will transition. The ones who have spent years coasting on process knowledge and positional authority will not. The flattening does not punish experience. It punishes rigidity.

From Training Departments to Learning Organizations

Peter Senge wrote about the “learning organization” in 1990, arguing that the ability to learn and translate learning into action would become a company’s ultimate competitive advantage. More than three decades later, most companies still have not built one. They have built training departments. The difference matters: training delivers information. Learning changes behavior.

In a flattened organization, the L&D function that matters is not the one producing the best course catalog. It is the one that figures out how to make a team learn faster and, more critically, how to make people genuinely enjoy learning. That second part is the hard part. Compliance-driven training cultures have spent years conditioning employees to associate “learning” with sitting through a module and passing a quiz. Undoing that conditioning requires deliberate effort.

Practically, this means L&D teams should shift focus in three areas:

1. Audit for curiosity, not just competency.

Most organizations measure whether employees can use new tools. Fewer examine whether they want to. L&D leaders should identify where intellectual curiosity is thriving and where it has gone dormant. The dormant zones are the real risk areas, because no amount of training will compensate for a team that has lost the desire to grow.

2. Design for speed of learning, not depth of content.

When organizations flatten and roles shift, the employees who survive the transition are the ones who can pick up a new domain in weeks rather than months. L&D teams should design programs that train the learning muscle itself: cross-functional rotations, rapid-prototyping challenges and scenario-based simulations that reward adaptability over mastery of any single tool.

3. Make the “beekeeper” transition explicit.

As AI absorbs routine tasks, mid-career professionals need help articulating what their beekeeper role looks like. That means exercising judgment, building relationships across organizational boundaries and providing contextual knowledge that AI cannot replicate. L&D programs should help employees practice directing AI rather than competing with it, turning reclaimed time into higher-value output rather than idle capacity.

The Make-or-Break Factor

Dorsey may be right that hierarchy was always just a bandwidth workaround. But removing it does not automatically create a better organization. What determines whether a flattened company thrives or collapses is its learning culture: whether the people who remain can adapt, grow and find meaning in continuous reinvention.

That is the make-or-break. Not the technology, not the org chart redesign, but whether a company has cultivated people who genuinely want to learn, or merely people who have learned to comply. L&D leaders who can build the former, who can inspire curiosity and make learning feel like a privilege rather than a mandate, will be the most important people in any flattening organization. The rest will find that the disappearing middle takes the learning function with it.