Using AI · Growth Library
Turn Real Examples into a Draft SOP with AI
Extract the repeated pattern, then test it on the next case
Try this
Choose several representative examples and remove information you should not share. Ask AI to separate inputs, steps, decisions, outputs, exceptions, and unresolved questions. Review the draft with the actual operator, then run one new case using only the SOP and revise where the procedure breaks.
Check-in: Could another competent person complete a fresh case from the checked SOP without relying on the hidden knowledge that created the examples?
Evidence: Practical guidance; no evidence-like claim is detected in the current source record.
Review note: Practical knowledge-capture workflow. Generated procedure text must be reviewed by the operator and tested on a fresh case before adoption.
AI can draft the manual nobody had time to write. The useful version starts from what the team actually does and survives contact with the next real case.
1. Choose representative examples
Include a normal case and at least one meaningful exception. Sanitise confidential information before using any external model.
2. Extract the procedure
Ask for inputs, ordered steps, decision points, outputs, exceptions, and unanswered questions. Require the model to distinguish what appears in the examples from what it is inferring.
3. Review with the operator
Have the person who performs the work correct missing branches, hidden prerequisites, unsafe shortcuts, and steps that exist only in theory.
4. Dry-run a fresh case
Use the SOP on a new example. Record every point where the operator needs knowledge that is absent from the document, then update the procedure.
Questions to consider
Why use examples instead of asking AI to write a generic SOP?
Examples expose the real sequence and exceptions in your workflow. A generic SOP can sound professional while describing a process nobody actually follows.
What if the examples disagree?
Do not force them into one clean story. Ask the model to surface the inconsistency and let the process owner decide whether it reflects a valid branch, an outdated practice, or an error.
Maintenance record
Review history
Current status: Reviewed. An editorial or evidence review is recorded and no later event changes that conclusion.
2026-09-15 · reviewed · Brali editorial agent Source: review-registry-using-ai-starters.json
A practical knowledge-capture workflow where generated documentation is validated against real operator behavior and a fresh case before adoption.
Browse the review ledger · Challenge or update this hack
Implementation observations can trigger a review, but they cannot change an evidence conclusion by themselves. Commercial relationships do not control review status or outcomes.
Continue from here
Choose the next useful path.
Article versions
Brali keeps substantive article history visible. Later reviews may refresh wording, sources, examples, or boundaries without silently replacing the record.
- Version 2026-09-15: first recorded publication in the migrated Brali corpus.
- Evidence review 2026-09-15: reviewed by Brali editorial agent; evidence state is
practical.
Canonical source record · Evidence state: practical.