Map a Decision Workflow

Outcome

A map shows the trigger, consequential choices, evidence, actors, handoffs, delays, errors, and beneficiary outcome.

Why It Matters

Automating a visible task can speed up the wrong constraint. Map the decision before selecting a model or tool.

Use When

Use for one live SME workflow with a named owner. Do not begin with a vendor shortlist.

Inputs

Bring one recent case, baseline time and quality, decision rights, current workaround, exceptions, privacy limits, and outcome evidence.

Recipe

1. Follow one case from trigger to beneficiary outcome.
2. Mark each decision, actor, input, output, wait, exception, and rework loop.
3. Distinguish judgment from transformation and retrieval work.
4. Locate the uncertainty whose reduction would most improve the outcome.
5. Choose one reversible augmentation boundary.

Human and AI Control

AI may organise observations and propose map gaps. People validate what actually happens and retain every consequential decision.

Output and Handoff

Return the map, baseline, bottleneck, candidate boundary, required evidence, owner, review point, and stop condition.

Evidence, Transfer, and Stop

Do not invent missing timings or outcomes. Transfer passes when another operator can trace a new case and recover the same decision points. Stop if mapping risks private data or worker surveillance.

Next

Design a model strategy for the bounded decision boundary.