Field note · Operations · 6 min read ·

Headwork: The Judgment Only You Can Provide

Everyone is asking what AI can do. The better question: what should humans still do?

The practical boundary around AI work is not tasks versus automation. It is judgment versus throughput. Some work needs speed, consistency, and memory. Some work needs an accountable person deciding what matters.

The teams that get value from AI do not try to remove people from every workflow. They separate the repeatable parts from the headwork: the parts where context, taste, risk, and responsibility still sit with a human operator.

The useful split

AI is strongest where the inputs are defined, the output can be checked, and the cost of a wrong draft is low. Human judgment is still required where tradeoffs are unresolved, accountability is high, or the organization has not agreed what good looks like.

That split is operational, not philosophical. It should show up in workflow design, review gates, escalation paths, and the way output is stored for reuse.

Designing for headwork

A good AI workflow preserves scarce human attention for the decisions only humans can make. It gathers the context, drafts the artifact, exposes uncertainty, and then asks for judgment at the right level of abstraction.

The result is not a person replaced by a tool. It is a person whose judgment is no longer buried under assembly work.

Begin with the diagnostic

A four-to-six-week assessment of what your firm knows.

What is leaking, what would compound if captured, delivered as a written readout. No commitment to what comes next.

Or read a sample readout

Further reading

Related field notes