Ask 'can a system handle this?' before 'can a person?' — system-first delegation prevents under-investment in automation

Why This Is a Rule

The default delegation instinct is person-first: "Who can I hand this to?" This skips the most scalable delegation target: systems. A person handling a delegated task uses their time on every instance. A system handling the same task uses zero ongoing time after the build investment. But systems are harder to set up than person-delegation, so the easier path (ask someone) gets chosen over the more leveraged path (build something).

The three-gate sequence corrects this bias: Gate 1 — Full automation: "Can a system (tool, script, template, automated process) handle this entirely?" If yes → build the system. One-time cost, permanent delegation. Gate 2 — Hybrid: "Can a system handle 80% with a person handling the 20% exceptions?" If yes → build the system for the routine cases and delegate only the exceptions to a person. Gate 3 — Full human judgment: "Does this genuinely require a person's contextual judgment for every instance?" Only then → delegate to a person.

The sequence forces system-first thinking. Most people jump to Gate 3, skipping the more leveraged Gates 1 and 2 entirely. The result is human attention consumed by tasks that don't require it.

When This Fires

• When deciding how to delegate a COULD DELEGATE task (Classify every task as ONLY ME, COULD DELEGATE, or SHOULD NOT EXIST — then eliminate or delegate everything outside ONLY ME)
• When assigning work to team members — check if a system could handle it first
• When recurring tasks consume human attention despite being routine and predictable
• When building delegation infrastructure (Distinguish ONLY ME by necessity from ONLY ME by default — tasks you never made delegable aren't genuinely non-delegable)

Common Failure Mode

Jumping to person-delegation: "I'll have the intern handle the weekly data report." Can a script handle 90% of the report with the intern reviewing edge cases? If yes, the intern's time is better spent on the 10% exceptions while the script handles the rest. Person-first delegation wastes human attention on automatable work.

The Protocol

(1) For each COULD DELEGATE task, run the three gates in order: Gate 1: "Can a system (script, template, automation, tool) handle this entirely without human involvement?" If yes → build or configure the system. Stop. Gate 2: "Can a system handle the routine cases (80%+) while a person handles exceptions?" If yes → build the system for routine cases, create an exception-handling protocol for the person. Stop. Gate 3: "Does every instance genuinely require human judgment?" If yes → delegate to a person with proper specifications (The competent stranger test: could someone with skills but zero context produce an acceptable result from your spec alone?). (2) Default assumption: most recurring tasks can be handled at Gate 1 or Gate 2. Reaching Gate 3 should be the exception, not the rule. (3) Review Gate 3 tasks quarterly: has the technology or tooling evolved to move them to Gate 1 or 2?