Remove friction
Use technology aggressively where the work is repetitive, rules-based, searchable, comparable, or mechanically transformable.
Technology is most valuable when it removes mechanical work without pretending that every consequential decision is mechanical.
Automate what repeats. Preserve the judgment that carries responsibility.
The idea
Efficiency and judgment are different kinds of work.
Automation can compress search, formatting, reconciliation, summarization, routing, comparison, and other repeated work. That creates time and attention. But the hardest institutional questions usually remain: what matters, what evidence is sufficient, what tradeoff is acceptable, who is accountable, and what should happen next.
The principle is therefore not anti-automation. It is a design rule for where automation should stop. The more consequential the decision, the clearer the human responsibility for purpose, interpretation, exception handling, ethics, and final authority should become.
Core principles
What to keep in view.
Keep accountability visible
A faster workflow should not make it harder to know who owns the decision, the exception, or the consequence.
Escalate ambiguity
Good systems recognize where confidence drops, context matters, or policy and values must be interpreted rather than merely executed.
Design for learning
Automation should create evidence about what works and where human intervention remains necessary, not simply conceal complexity behind a smoother interface.
Boundaries
What the idea does—and does not—claim.
Not human exceptionalism
Machines may outperform people in many analytical tasks. The point is responsibility, not preserving work for its own sake.
Not automation avoidance
Keeping humans in every loop can create delay, inconsistency, and false comfort. The human role should be purposeful.
Not a fixed boundary
As systems improve, the right allocation of work can change. Governance should be revisable as evidence changes.
Put it to work
Where the idea becomes practical.
AI-assisted knowledge work
Let AI retrieve, compare, draft, summarize, and test coherence while people retain responsibility for purpose, truth, meaning, and consequential choice.
Operations
Automate repeatable routing and reconciliation, then route exceptions and ambiguous cases to accountable people.
Governance
Use tools to surface evidence and patterns without converting fiduciary, ethical, or policy judgment into a hidden scoring function.
Leadership
Free people from clerical repetition so more capacity can move toward judgment, relationships, problem framing, and learning.
Go deeper