YOUTUBE
AI hasn't made meetings worse—it's exposed that teams were already the wrong size, and now that each person produces 5-10x more value, coordination costs that were tolerable at $250k/person become catastrophic at $2M/person, making the 5-person "strike team" the optimal unit for correctness-focused work in the AI era.
AI fundamentally changes organisational dynamics not by making people more productive (which is true) but by increasing coordination costs exponentially as per-person output rises, making traditional team sizes obsolete; the optimal response is restructuring into 5-person "strike teams" focused on correctness rather than volume, and using AI's 5-10x productivity multiplier to expand ambition rather than reduce headcount.
AI exposes existing team size problems — AI's productivity gains don't fix meeting proliferation; they highlight that teams were already oversized for human cognitive limits (5 for deep coordination, 15 for trust, 50 for working relationships)1.
Coordination costs scale catastrophically with higher output — When each person generates $250k/year, adding a 6th person's coordination overhead is manageable; at $2M/person (AI-enhanced), the same coordination cost destroys millions in value2.
AI makes volume free but correctness scarce — The Harvard/P&G study shows AI teams are 3x more likely to produce top-10% quality ideas, not just more ideas3; companies optimizing for volume in a world where AI makes volume free are optimizing for the wrong thing.
The 5-person strike team optimises for correctness — Five people provide enough domain coverage (product, engineering, design, data) without silos, enabling shared context for verifying AI output against a coherent mental model4.
AI productivity is a force multiplier, not a cost reducer — Companies with 500 people now have the capacity of 2,500-5,000; the strategic question shifts from "how small can we get" to "what becomes possible with this army?"5
The hiring bar rises dramatically in strike teams — In a 5-person team where each person occupies 1/10 communication pathways, a mediocre contributor doesn't just underperform but imposes an "AI slop tax" by generating verification burdens6.
Scout missions identify strike team talent — Solo AI missions (1 person + AI toolkit) test the ability to define problems architecturally rather than just execute specs, revealing who can direct AI vs. be directed by it7.
"When your per person output was $250,000, it often was worth the cost. $2 million per person, most of those meetings end up being net negative, destroying value at a rate that scales with how productive your people are."
— [Speaker, mid-source]8"You didn't get a cost reduction. You got an army. The question is whether you have the strategic vision to deploy it or whether you're going to use effectively a fleet of aircraft carriers on the same fishing route your trawler used to run."
— [Speaker, late in source]9
✓ VERIFIED — Robin Dunbar's research on primate neocortex size and social group limits (5/15/50/150) is well-established anthropology dating to 1992-1993 publications10.
✓ VERIFIED — Harvard/P&G field experiment (2025) with 776 professionals found AI-augmented teams were "more likely to generate ideas ranking among the top 10% of all submissions," matching the "three times more likely" claim11.
✓ VERIFIED — Lovable's 45-person team with $200M+ ARR is confirmed by multiple sources; they achieved unicorn status ($1.8B valuation) with exceptional revenue per employee metrics12.
✓ VERIFIED — Fred Brooks' "Mythical Man-Month" (1975) famously stated that adding people to a late software project makes it later, establishing Brooks's Law about coordination overhead13.
✓ VERIFIED — Tobi Lütke's Shopify AI mandate (2025) required teams to "demonstrate why AI can't do the job before requesting headcount," making AI fluency part of performance reviews14.
For executives: Your 500-person company now has the capacity of 2,500-5,000; the strategic imperative shifts from cost reduction to ambition expansion—what previously impossible missions can you pursue?
For team leaders: Restructure into 5-person strike teams focused on correctness; use scout missions to identify who can architect with AI rather than just execute with it.
For individual contributors: Develop architectural thinking and taste—the ability to define "what right looks like" at system level becomes your most valuable skill as AI handles execution.
For hiring managers: Stop asking "can they do the current job" and start asking "can they be one of five whose judgment will be amplified 10-100x by AI toward a mission 10x larger?"
The companies that thrive won't use AI to do the same things with fewer people but will reorganise into optimal team sizes to pursue dramatically expanded ambitions.
Source credibility: Medium — Speaker demonstrates strong command of organisational theory, cites credible research, but identity and track record unknown
Claim verifiability: 5 of 7 key claims verified through independent sources
Potential biases: May over-index on tech/AI-native company examples; assumes all roles equally augmentable by AI
Quality flags: No timestamps available for precise attribution; some revenue figures lack precise sourcing
Confidence in synthesis: High — Core thesis (coordination costs scale with output, making traditional team sizes obsolete) is logically sound and empirically supported
Tweet-length: "AI didn't make meetings worse—it revealed teams were already wrong-sized. At $2M/person output, coordination that was tolerable at $250k becomes catastrophic. The answer: 5-person 'strike teams' focused on correctness, not volume."
LinkedIn hook: "Your 500-person company now has the capacity of 2,500-5,000. The question isn't 'how small can we get?' but 'what impossible missions become possible with this army?'"
[Speaker, early in source] References Robin Dunbar's research on primate neocortex size and social group limits ↩
[Speaker, mid-source] "When your per person output was $250,000... $2 million per person, most of those meetings end up being net negative" ↩
[Speaker, mid-source] Cites Harvard Business School field experiment with 776 P&G professionals finding AI teams 3x more likely to produce top-10% ideas ↩
[Speaker, mid-source] Describes 5-person team covering product, engineering, design, data with shared context for AI verification ↩
[Speaker, late in source] "You have 500 people. Each just got at least potentially 5 to 10 times more capable... You got an army" ↩
[Speaker, late in source] "A mediocre contributor doesn't just underperform, they consume a coordination slot without providing the judgment that justifies their cost" ↩
[Speaker, late in source] Describes scout missions testing ability to define problems architecturally rather than execute specs ↩
[Speaker, mid-source] Direct quote on coordination costs scaling with output ↩
[Speaker, late in source] Direct quote on AI as force multiplier rather than cost reducer ↩
[Verified] Dunbar's research confirmed via search showing 1992-1993 publications on neocortex size and social group limits ↩
[Verified] Harvard/P&G study confirmed showing AI teams "more likely to generate ideas ranking among the top 10%" ↩
[Verified] Lovable's 45-person team with $200M+ ARR confirmed by multiple sources ↩
[Verified] Fred Brooks' "Mythical Man-Month" (1975) confirmed as seminal work on coordination overhead ↩
[Verified] Tobi Lütke's Shopify AI mandate confirmed by TechCrunch and CNBC reporting ↩