AI Business Process Review Workflow

5 steps to a reviewable draft

1. 01

   Reconstruct the current process

   Record the trigger, steps, queues, handoffs, exceptions and finish condition from recent work.

   Result: A current-state map based on what people actually did.

2. 02

   Add operating evidence

   Attach available wait times, rework counts, missed handoffs and examples where the process worked normally.

   Result: A small evidence set that separates common friction from one-off incidents.

3. 03

   Generate testable hypotheses

   Ask for possible bottlenecks and the evidence needed to confirm or reject each one.

   Result: A shortlist of causes that can be checked rather than accepted on sight.

   Your prompt

   Copy prompt

   Review this business process: {{process}}. Recent examples and timings: {{evidence}}. Exceptions and rework: {{exceptions}}. Owner constraints: {{constraints}}. Return observed friction, up to four bottleneck hypotheses, the evidence for and against each, validation questions and one small reversible test. Label every unconfirmed cause [HYPOTHESIS]. Do not claim causation, savings, capacity gains or employee performance without supplied evidence.

   Example input

   Invoice approval takes 2-8 days. Three recent invoices waited for missing purchase details; two waited for owner approval.

   Example output

   [HYPOTHESIS] Intake fields cause avoidable rework. Test: require purchase details on the next ten requests and compare missing-field returns.

4. 04

   Review with the people doing the work

   Ask operators which observations are wrong, what the data missed and what a proposed test might disrupt.

   Result: One accepted hypothesis and safe test boundary.

5. 05

   Record the test

   Name the owner, baseline, change, review date and rollback condition before altering the process.

   Result: A process experiment with a result the team can compare.