Technical Context
Instead of reading summaries, I went to the source: a March note from Block written by Dorsey and Roelof Botha. The core idea isn't the buzzword 'AI,' but a very specific company restructuring. They are moving coordination from a human hierarchy to a central AI system that maintains a real-time business model and syncs context between people.
This is where it clicked for me. Usually, companies buy another copilot and call it an 'AI transformation.' But this idea is more radical: AI doesn't just assist the organizational structure; it becomes its backbone.
Roles have also been redesigned unsentimentally. Deep specialists become ICs (Individual Contributors) and get context directly, without a managerial layer. Product and initiative owners become DRIs (Directly Responsible Individuals) and truly own their tasks, including access to resources through the shared system.
Traditional middle management is being cut. Some are becoming player-coaches—not administrators who track statuses, but people who both contribute directly and develop the team. Everything that used to be handled by the 'call, forward, sync, clarify, escalate' layer is, by design, moving to the AI loop.
On paper, this sounds almost like a mini-AGI for the company, and Block itself frames it this way. But I wouldn't romanticize it. There are no public details yet about the tech stack, model quality, access controls, decision auditing, or how they handle errors in the shared business memory.
There's also important context: in early 2024, Block laid off around a thousand employees. That's why I read this not as abstract management philosophy, but as a high-stakes operational bet: a smaller team, higher context density, and a higher cost of error.
What This Changes for Business and Automation
Realistically, Dorsey is demonstrating not just 'another AI implementation case,' but a new model for organizational design. Here, AI solves the most expensive problem in a growing company: transferring context between functions. Not generating text or powering a website chatbot, but pure coordination.
For fast-moving teams, this is a powerful move. When a DRI can quickly assemble the necessary IC resources and a specialist gets full context without five sync meetings, the decision cycle truly shrinks. I see this in my clients' cases too: as soon as we remove manual knowledge routing, AI automation starts generating revenue, not just LinkedIn likes.
Who wins? Companies with strong individual contributors, good data discipline, and a willingness to define processes systematically. For them, an AI architecture can become a control center, not just a showcase. This is especially effective in product companies, fintech, SaaS, and operationally heavy businesses where people drown in status updates.
Who loses? Organizations where the managerial layer relies on personal agreements, politics, and the institutional memory of 'old-timers.' If the business doesn't exist in structured data, a central AI system won't synchronize it; it will just hallucinate beautifully. And that will be painful.
There's a second, more human risk. Not every strong manager will become a player-coach, and not every expert wants to be an IC with increased autonomy. This kind of restructuring breaks career ladders, motivation, and the familiar habit of hiding weak decisions behind collective responsibility.
So, I wouldn't copy Block from a slide deck. I'd adopt the principle: first, map the knowledge; then, define access rights; then, implement a DRI model; and only then, integrate AI into the operational loop. Otherwise, you'll end up with expensive chaos under a fancy name.
At Nahornyi AI Lab, this is exactly the layer we work on. We don't just bolt on a model; we build an architecture of AI solutions around real processes, task owners, and data flows. This is where theory ends and engineering, access controls, orchestration, and very pragmatic questions of accountability begin.
This analysis was written by me, Vadym Nahornyi of Nahornyi AI Lab. I work with AI automation not in presentations, but in live systems where you need to unite data, roles, and actions into a single working mechanism. If you want to discuss your project and see where AI can truly make an impact, contact me, and we'll explore it together.