Principles for AI Delegation

Table of content

The goal isn’t to automate everything. It’s to automate the right things.

The delegation spectrum

ALWAYS HUMAN          DELEGATE + REVIEW          FULLY AUTOMATE
     ←─────────────────────────────────────────────────→

Strategic decisions    Complex research          Data entry
Creative vision        Draft writing             Formatting
Relationship calls     Scheduling                Reminders
Value judgments        Code review               Status updates
Final approval         Analysis                  Triage

What to delegate

Confidence	Tasks	Your role
High	Info retrieval, formatting, scheduling, data transformation	Trust output
Medium	Draft writing, prioritization, research, code generation	Review and edit
Keep human	Strategy, relationships, creative choices, ethics	Do yourself

The review gradient

Risk level	How to review	Example
Low	Glance, trust unless off	Formatted document
Medium	Read completely, check facts	Email draft
High	Verify claims, check reasoning	Strategic recommendation

Match review depth to stakes.

The learning loop

1. Delegate task
2. Review output
3. Correct errors (explicitly, with context)
4. AI learns preferences
5. Next time is better

Only works if you actually correct mistakes — don’t just fix silently.

Common mistakes

Mistake	Signs	Fix
Over-delegation	Don’t understand AI’s choices, stakes higher than realized	Keep more human judgment
Under-delegation	Repetitive work, no judgment needed, process over substance	Delegate more
Unclear delegation	AI misses the point, more fixing than doing	Better prompts

Decision framework

For any task, ask:

Question	If yes →	If no →
Downside of AI error is minor?	Delegate freely	Review or keep human
I’d notice if AI got it wrong?	Delegate freely	Be careful
Easy to fix AI’s output?	Delegate freely	Keep human

The meta-principle

Delegate the process, keep the judgment.

AI handles	You handle
Gathering information	Choosing between options
Structuring options	Defining what matters
Executing decisions	Final calls

This leverages AI’s strengths (speed, thoroughness, consistency) while preserving yours (values, creativity, accountability).

Next: Getting Started Guide