THE SHORT ANSWER
List the tasks that produce your role's outcomes, then assess frequency, variation, data, judgment, interaction and consequence. Classify each as a candidate to automate, augment, keep human or redesign, and revisit the map as tools and responsibilities change.
Start with actual work
Calendar entries and job descriptions often miss informal coordination, corrections and exceptions. Observe a typical week and write tasks as actions with outputs: reconcile invoice exceptions, prepare a client brief, approve a campaign change.
Evidence & context: GOV.UK Service Manual
Assess conditions before choosing a label
| Question | Why it matters |
|---|---|
| How repeatable is it? | Stable patterns are easier to standardize |
| How much variation exists? | Variation raises context and exception needs |
| What judgment is required? | Ambiguity and consequence need accountable review |
| What human interaction matters? | Trust, negotiation and care may be part of the outcome |
| What data or access is needed? | Privacy and permission may constrain the design |
Classify a next experiment
- AUTOMATE: stable execution with clear rules and recovery.
- AUGMENT: AI helps while a person interprets and decides.
- KEEP HUMAN: interaction, consequence or context dominates.
- REDESIGN: remove, simplify or change the workflow before adding AI.
A label is provisional. The jagged capability frontier and changing tools require testing on representative work.
Evidence & context: Harvard Business School
Turn task change into a learning decision
For each augmented or redesigned task, name the capability the human now needs: better problem framing, evaluation, exception handling, data literacy, stakeholder communication or system supervision. This produces a focused reskilling plan instead of ‘learn AI’.
Sources & further reading
- Generative AI and Jobs: A Refined Global Index of Occupational Exposure
International Labour Organization and NASK. A 2025 working paper combining task-level data for nearly 30,000 occupational tasks, expert validation, model-assisted scoring and harmonized employment data. Exposure indicates potential task transformation; it is not a forecast that a job will disappear.
- Navigating the Jagged Technological Frontier
Harvard Business School. A field experiment with 758 BCG consultants performing selected knowledge-work tasks. AI effects differed depending on whether a task was inside or outside the model's capability frontier; the sample and tasks limit broader inference.
- Map and understand a user's whole problem
GOV.UK Service Manual. Public-service guidance on mapping journeys, collaborating across boundaries and locating pain points. The module adapts the method to business workflows.
Examples and exercises are illustrative unless attributed to a source. No independent expert review is claimed.
A correction, a counterexample or an experience worth sharing?
Join the conversation ↗