Most important take away

Companies rushing to flatten management with AI are throwing out load-bearing human functions alongside the automatable ones. The three bundled jobs of management — information routing, sense making, and accountability/feedback — must each be deliberately addressed when restructuring; eliminating managers without a plan for all three creates culture strain, burnout, and attrition regardless of how fast you ship.

Summary

Actionable Insight 1: Understand the three bundled jobs of management before cutting. Managers perform three distinct functions: (1) information routing (who needs what, when), (2) sense making (interpreting signals and filtering noise), and (3) accountability and feedback (ownership, coaching, and performance). AI can handle routing today, can assist with sense making, but cannot replicate the human depth required for accountability and long-term ownership. Before removing management layers, decompose these functions and assign each one explicitly.

Actionable Insight 2: If you are a manager, shift your visible value toward what AI cannot do. If your role is mostly information routing, you are vulnerable. Deliberately demonstrate your sense-making ability — pattern recognition across projects, translating ambiguous signals into decisions. Show your coaching and accountability contributions. Become a player-coach (IC work plus people development). Use AI to augment feedback gathering and data synthesis, then visibly apply human judgment on top.

Actionable Insight 3: Three real-world models are being tested — study them.

• Kimi (Moonshot AI, 300 people): Agents handle routing, five co-founders handle all sense making for ~50 direct reports each, accountability is left to self-reflection. Result: extraordinary speed but significant human cost — burnout, crying in meetings, senior hires leaving, a feeling of “weightlessness” that becomes anxiety.
• Block (Jack Dorsey): AI-powered “world model” handles routing, Directly Responsible Individuals (DRIs) own cross-cutting problems for 90-day rotations with full authority (sense making), and player-coaches handle accountability and people development. This is the most structurally innovative approach but has not been fully tested yet.
• Meta (Zuckerberg): Compressed the management bundle rather than decomposing it — fewer managers with wider spans, AI assists routing within existing hierarchy, accountability is intensified (bottom 5% cut). Shipping faster with strong stock performance, but internal reports suggest unsustainable pressure and burnout.

Actionable Insight 4: Start automation with information routing. If you lead a team or organization, automate information routing first. It is the lowest-risk, highest-return management function to hand to AI. Synthesizing status updates, cascading policy changes, aggregating team reports — these are solved problems for AI today.

Actionable Insight 5: Name owners with explicit authority. Jack Dorsey’s DRI model highlights the power of naming a single person responsible for a cross-cutting problem with real authority and an expiration date. This prevents permanent middle-management bloat while preserving human sense making close to the problem. Consider adopting time-bounded ownership rotations in your organization.

Actionable Insight 6: For individual contributors — find a good manager. The single largest predictor of whether someone thrives at work is their relationship with their manager. In a reshuffling landscape, invest time in finding leaders who understand the decomposed management model and can provide genuine coaching, sense making, and accountability. A good manager is worth their weight in gold.

Career Advice:

• If you are a manager whose role is mostly routing, retool immediately — that function is being automated.
• Demonstrate IC capability alongside management skills; player-coaches are the model gaining traction.
• Learn to use AI to gather broader feedback data and market signals, then layer human interpretation on top — this makes you indispensable.
• If you are in the C-suite, decompose management roles to first principles before making cuts. Companies that compress without decomposing risk losing the human functions that drive retention and decision quality.

Chapter Summaries

The Three Bundled Jobs of Management Managers historically perform three functions: information routing (logistics of who knows what, when), sense making (filtering signal from noise and interpreting ambiguous information), and accountability/feedback (ownership, coaching, performance management). These have been bundled into the manager role for centuries, tracing back to Roman centurions and Prussian military staff.

Where AI Fits in Each Function AI excels at routing — synthesizing updates and distributing information is essentially solved. Sense making is partially automatable; AI can process raw data but lacks the deep domain context and human judgment needed for reliable interpretation. Accountability and feedback remain fundamentally human; long-running ownership, emotional investment, and nuanced coaching cannot yet be simulated by agents.

Case Study: Kimi (Moonshot AI) A flat, 300-person AI-native company with zero hierarchy, zero titles, zero OKRs. Agents handle routing brilliantly (a PM launches three agents before lunch and has code shipping by midday). Five co-founders carry all sense making for ~50 people each. Accountability is absent — former employees report not knowing what to work on, and the culture produces both extreme speed and real human casualties including burnout and senior departures.

Case Study: Block (Jack Dorsey) Dorsey proposes a deliberate decomposition: an AI “world model” for routing, DRIs with 90-day rotating ownership for sense making, and player-coaches for accountability. The sharpest structural innovation is assigning sense making to the person closest to the problem with explicit authority and an expiration date. This model has not yet been fully deployed; Block recently lost half its workforce and is in a cultural reset.

Case Study: Meta (Zuckerberg) Meta compressed rather than decomposed management — fewer managers with wider spans, AI-assisted routing, and intensified accountability (bottom 5% performance cuts). Stock performance is strong and shipping speed has improved, but internal reports suggest unsustainable pressure and potential for a revolving door of talent.

Implications for the Future of Management The fundamental question is whether companies that decompose management functions deliberately will outperform those that simply compress them. The ability of leaders to specifically imagine where AI fits in their organization is a strong positive indicator for AI adoption. This is one of the largest experiments in management theory in 2,000 years, and the outcomes will shape how all organizations adopt AI.