You will leave with a clear personal workflow, a test log, and a version you can reuse without losing control.
Module 03 · Playbook
Playbook: from judgment to Mini-SOP
You will leave with a clear personal workflow, a test log, and a version you can reuse without losing control.
- Practice
- Choose a task, observe your practice, compare sources, co-create the Mini-SOP, and test one case under human review.
- Transfer
- Repeat the workflow on a second case and change only what the log supports.
Turn one real task into a way of working you can repeat.
When every delivery depends on memory and improvisation, you repeat decisions, carry friction, and cannot tell what AI improved.
Choose a task, observe your practice, compare sources, co-create the Mini-SOP, and test one case under human review.
A Mini-SOP with input, steps, criteria, human review, output, exceptions, and a real-case test log.
Full index14
00
Starting point
Starting point
Make your judgment visible, use AI to prepare and compare, and finish with a Mini-SOP you have tested yourself.
You will leave with a clear personal workflow, a test log, and a version you can reuse without losing control.
- Challenge
When every delivery depends on memory and improvisation, you repeat decisions, carry friction, and cannot tell what AI improved.
- Practice
Choose a task, observe your practice, compare sources, co-create the Mini-SOP, and test one case under human review.
- Evidence
A Mini-SOP with input, steps, criteria, human review, output, exceptions, and a real-case test log.
01
Chapter 01
Amplification boundary
Principle: Judgment before assistance
AI prepares options; a person retains purpose, boundaries, and decision.
Go deeperHow to apply it
How to apply it
- Name the human decision.
- Define what AI prepares.
- Set a stop.
Checklist
- Name the human decision.
- Write what you decide.
- Decision rule
- Without an owner, it is not ready.
- Example
- Personal synthetic example: AI organizes; you decide.
- Antipattern
- Automatic approval because it “looks right”.
- Expected evidence
- Personal boundary.
- Limit
- Do not delegate explainable decisions.
02
Chapter 02
Choose a useful workflow
Principle: Small, bounded workflow
Start with a frequent task whose failure can be contained.
Go deeperHow to apply it
How to apply it
- Bound start and finish.
- Choose a case.
- State risk.
Checklist
- Bound start and finish.
- Choose your own task.
- Decision rule
- Choose for usefulness and evidence.
- Example
- Personal synthetic example: Non-sensitive weekly report.
- Antipattern
- Testing a critical process first.
- Expected evidence
- Workflow brief.
- Limit
- Protect data and consequences.
03
Chapter 03
Friction map
Principle: Friction before solution
Observe waiting, rework, handoffs, and repeated decisions.
Go deeperHow to apply it
How to apply it
- Reconstruct a case.
- Mark friction.
- Ask about causes.
Checklist
- Reconstruct a case.
- Accept observed friction.
- Decision rule
- Cause remains a hypothesis.
- Example
- Personal synthetic example: Repeated search in each delivery.
- Antipattern
- Blaming from one observation.
- Expected evidence
- Friction map.
- Limit
- Do not turn impressions into causes.
04
Chapter 04
Input, output, and done
Principle: Done is checkable
Inputs, outputs, and criteria prevent fluency being confused with quality.
Go deeperHow to apply it
How to apply it
- Name user.
- Write criteria.
- State rejection.
Checklist
- Name user.
- Set your done criteria.
- Decision rule
- Criteria support acceptance or correction.
- Example
- Personal synthetic example: Two sources and one decision.
- Antipattern
- Using “make it good”.
- Expected evidence
- Definition of done.
- Limit
- Do not close without checks.
05
Chapter 05
Source comparison
Principle: Sources with context
A practice guides when provenance, conditions, and limits remain visible.
Go deeperHow to apply it
How to apply it
- Ask from friction.
- Compare practices.
- Record adaptation.
Checklist
- Ask from friction.
- Adapt to your boundaries.
- Decision rule
- Cite and explain fit.
- Example
- Personal synthetic example: Adapted checklist.
- Antipattern
- Accepting unsourced lists.
- Expected evidence
- Compared dossier.
- Limit
- Do not use unknown sources.
06
Chapter 06
Mini-SOP anatomy
Principle: Concise, complete Mini-SOP
Connect purpose, input, steps, criteria, review, output, and exception.
Go deeperHow to apply it
How to apply it
- Write minimum sequence.
- Add manual path.
- Record version.
Checklist
- Write minimum sequence.
- Keep a usable Mini-SOP.
- Decision rule
- Every step governs action or decision.
- Example
- Personal synthetic example: One-page Mini-SOP.
- Antipattern
- Documenting without who decides.
- Expected evidence
- Mini-SOP v0.1.
- Limit
- Do not hide exceptions.
07
Chapter 07
AI co-creation
Principle: Co-creation, not delegation
AI proposes; humans compare, correct, and authorize.
Go deeperHow to apply it
How to apply it
- Provide context.
- Request traceable draft.
- Review assumptions.
Checklist
- Provide context.
- Approve every change.
- Decision rule
- A suggestion is not a decision.
- Example
- Personal synthetic example: Reject an assumption.
- Antipattern
- Approving without comparison.
- Expected evidence
- Change log.
- Limit
- Do not accept unjustified changes.
08
Chapter 08
Human-in-the-loop control
Principle: Review at the right point
The human gate occurs before the effect and leaves a decision.
Go deeperHow to apply it
How to apply it
- Locate gate.
- Assign reviewer.
- Define options.
Checklist
- Locate gate.
- Stop when the gate fails.
- Decision rule
- Review after the effect does not prevent.
- Example
- Personal synthetic example: Review citations.
- Antipattern
- Symbolic review without authority.
- Expected evidence
- Personal gate.
- Limit
- Do not continue after gate failure.
09
Chapter 09
Controlled test
Principle: A test that enables learning
The pilot retains inputs, failures, and decisions; AAR guides adjustments.
Go deeperHow to apply it
How to apply it
- Run a case.
- Compare criteria.
- Authorize changes.
Checklist
- Run a case.
- Decide after AAR.
- Decision rule
- An intact failure also teaches.
- Example
- Personal synthetic example: Missing source stops test.
- Antipattern
- Polishing the log.
- Expected evidence
- Log and AAR.
- Limit
- Do not change many variables.
10
Chapter 10
Version and transfer
Principle: Transfer before adoption
Another case or person must run, stop, and escalate.
Go deeperHow to apply it
How to apply it
- Test beyond first case.
- Record help.
- Decide version.
Checklist
- Test beyond first case.
- Keep only with second evidence.
- Decision rule
- One test does not authorize broad use.
- Example
- Personal synthetic example: Second report.
- Antipattern
- Calling one test a standard.
- Expected evidence
- Case comparison.
- Limit
- Do not generalize without limits.
11
Glossary
Glossary
- Friction map
- A record of waiting, rework, handoffs, and repeated decisions.
- Mini-SOP
- A concise, versioned procedure with steps, criteria, and exceptions.
- Human gate
- A point before an effect where a person decides.
- AAR
- A review of expectation, result, evidence, and adjustment.
- PASA
- Plan, accelerate, systematize, and amplify what is stable.
12
FAQ
Frequently asked questions
Must I automate?
No. First expose the workflow and keep a manual path.
How many cases?
Two checks support learning, not broad claims.
Can AI draft it?
It prepares a draft; you compare and approve.
What if a source is missing?
Stop the dependent part and record the gap.
Repeat the workflow on a second case and change only what the log supports.
Make your judgment visible, use AI to prepare and compare, and finish with a Mini-SOP you have tested yourself.
Checklist
Amplified work- I explain friction with evidence.
- I retain judgment and a manual path.
- Test and AAR justify changes.
- The second case bounds transfer.