Claim
When a team has access to a strong AI substrate, deliberately under-staff each project — one person plus the substrate beats a fully-staffed team — because constraint forces clarity and lets the substrate absorb the rest.
Mechanism
Adding humans to a workstream the substrate can do creates coordination overhead, slower decisions, and weaker individual ownership. A single under-resourced operator with strong tooling has to clarify scope to ship at all; that forced clarification is where the productivity comes from. The substrate eats the throughput gap.
Conditions
Holds when:
- The substrate is genuinely capable for the task domain (Claude Code for engineering; this is now plausible for many marketing/PM workstreams).
- The single operator has the judgment to scope and prioritise.
- The org culture rewards intrinsic motivation, not headcount-as-signal-of-importance.
Fails when:
- The work is genuinely beyond the substrate's reach (novel research, deep multi-stakeholder coordination).
- The single operator is junior and needs explicit teammates for craft growth.
- Organisational politics treat headcount as currency for promotion.
Evidence
"There's this interesting thing that happens when you underfund everything a little bit — people are forced to clarify, and the way they ship really quickly is just intrinsic motivation."
"Productivity per engineer increased 200% while we 4x'd the team."
Anthropic measured both more people and more output per person. Boris ships 10–30 PRs/day with ~5 Claude Code agents in parallel and is still one of the most prolific committers on his team.
— Boris Cherny on Lenny's Podcast, 2026-04-27
Signals
- Time-to-first-decision on a new workstream drops; a single named owner is set within hours, not weeks.
- Engineers / PMMs report that token spend rose but headcount stayed flat or grew with output.
- Projects that historically required 4 people now ship with 1 + substrate.
- Revenue or output per employee rises while team size grows (compound, not substitution).
Counter-evidence
Operators with no substrate access, or in regulated domains where every artifact must be co-reviewed, will hit a quality wall fast. Underfunding without compensating substrate is just the old "do more with less" trap. Boris is explicit that the win comes from substrate + token budget, not constraint alone.
Cross-references
- Agents work when treated as a team, not a single super-tool — the substrate side of the same equation
- The unlock for AI agent productivity is management skill, not technical skill — the operating-discipline prerequisite