Jack Dorsey and the Block team published a massive manifesto about how they're restructuring internal processes to fit the AI reality. You probably remember how they recently laid off 4,000 people. That move locked in a headcount cap — and now they're structurally adapting to do more with less.
In Dorsey's view, classic hierarchy is obsolete. Middle managers used to act as transmitters: collecting team status updates, passing them up the chain, and pushing business context back down. A slow process with a lot of data loss along the way. Now an AI system can hold up-to-date information about the entire company and instantly surface the right context for any employee.
Quoting Dorsey: "Block is tearing down the multi-layered management structure and keeping just three role types":
🔸 Individual contributors (IC) — specialists who do the hands-on work. They get business context from AI, not through a chain of managers.
🔸 DRI (Directly responsible individuals) — people accountable for specific complex problems who rally the right resources around them.
🔸 Player-coaches — no more pure managers. Leaders must be practitioners themselves (writing code, designing, doing research) while also mentoring talent.
He then explains how to adapt to all of this.
What businesses / founders should do:
Stop hiring people to be "information relays." If someone's job is pinging others in Slack and rolling up statuses into a spreadsheet — it's time to automate that.
Build internal knowledge bases that LLMs can easily read. AI should become the main onboarding tool and context keeper for teams.
Flatten the structure. Speed up decision-making by giving individual contributors direct access to data.
What specialists / employees should do:
Get back to hands-on work. If you've spent the last five years managing development without writing a single line of code yourself — you're at risk. The market is moving toward the player-coach model. Hard skills matter more than pure task management ability.
Learn to work autonomously. In the AI-native future, there's no lead standing over you breaking down the task. You get a high-level goal from leadership, pull all the context you need from the company's AI assistant, and deliver results.
Build your DRI muscle — the ability to own the outcome of a complex task, independently assembling the right AI tools and people to get it done.
Personally, I'm building a Personal OS for exactly this. In one of my next posts I'll go into detail about how I'm combining personal and work contexts and how it all gets automatically distributed across databases (Obsidian, Notion, GitHub).