← All Concepts

Canon

Complementarity

/ˌkɒmplɪmɛnˈtærɪti/

From Latincomplementum: "that which fills up"

Published January 8, 2026

Definition

The principle that human and AI capabilities complete rather than replace each other. Neither is sufficient alone; together they form a whole greater than the sum of parts. The human provides judgment, context, and purpose; the AI provides scale, consistency, and tireless execution.

Origin

The term derives from Niels Bohr's principle in quantum mechanics, where wave and particle descriptions are both necessary but mutually exclusive views of the same phenomenon. CREATE SOMETHING adapts this to human-AI collaboration: different modes of intelligence that cannot be reduced to each other but together describe the complete picture.

In the CREATE SOMETHING context, complementarity emerged from practical experience with Claude Code—discovering that the most effective work happens when human and AI each do what they do best, with clear handoff points rather than blurred responsibilities.

In Practice

The Complementarity Principle manifests in how CREATE SOMETHING divides work between human and AI:

Human (WezTerm)

  • Monitor logs
  • Verify production
  • Debug edge cases
  • Interactive sessions
  • Observe and decide
  • Provide context and judgment
  • Set direction and purpose

AI (Claude Code)

  • Write code
  • Deploy code
  • Run migrations
  • Execute tests
  • Plan and execute
  • Scale repetitive tasks
  • Maintain consistency

The boundary is not rigid but principled: handoff occurs where one mode of intelligence reaches its limit and the other excels.

Relationship to Zuhandenheit

Complementarity enables Zuhandenheit (ready-to-hand). When human and AI work in their proper domains, the tool recedes. The human doesn't think about the AI; the AI doesn't demand attention. They simply work together.

When complementarity breaks down—when the AI attempts judgment it cannot make, or the human micromanages execution—the tool becomes present-at-hand. We notice the partnership instead of the work.

Anti-Patterns

What Complementarity is not:

  • Replacement — AI replacing human judgment, or humans doing what AI does better. Complementarity is completion, not substitution.
  • Delegation without oversight — The human remains responsible. Complementarity is partnership, not abdication.
  • 50/50 split — The division is by capability, not by quantity. Some tasks are 90% AI; others are 90% human.
  • Static boundaries — As AI capabilities grow, the boundary shifts. What required human judgment yesterday may be AI-appropriate today.

The Complementarity Test

When deciding who should do a task, ask:

"Does this require judgment, or execution?"

Judgment—weighing tradeoffs, understanding context, deciding what matters—belongs to humans. Execution—implementing decisions consistently at scale—belongs to AI. The boundary is where one transforms into the other.

In the Hermeneutic Circle

Complementarity connects to the broader CREATE SOMETHING system:

  • .agency applies complementarity to client work—AI handles implementation, humans handle relationships and strategy.
  • .io uses complementarity in research—AI gathers and synthesizes, humans interpret and apply.
  • .space explores complementarity—experiments test where the boundary lies.
  • .ltd defines complementarity—Canon documents the principle so it can be applied consistently.

References