Knowledge Systems

Turn an Expert Interview into an SOP

Convert an expert interview into a source-attributed, reviewable SOP while preserving uncertainty, conflicts, and human ownership.

Quick facts

Best for
Owners · Founder-operators
Prompt type
Quality
Expected result
A source-attributed, reviewable SOP with evidence states, numbered steps, explicit Unknowns, an Unresolved conflict register, and a human approval checklist.
Time saved
45-60 minutes per SOP draft
Required inputs
INTERVIEW TRANSCRIPT · SOURCE OWNER · SYSTEM OR LOCATION · SENSITIVITY · AUTHORITY · LAST-REVIEWED DATE · SOP TEMPLATE
Works with
ChatGPT + Claude

What this prompt does

  • Structures supplied expert knowledge into a step-by-step draft with source attribution and evidence states.
  • Separates verified instructions from Unknown details and unresolved business-rule conflicts for owner review.

When to use

  • After a subject-matter interview when the team needs a traceable first SOP draft, not a silently invented process.

When not to use

  • Do not use it to publish an SOP, replace the canonical source, decide which conflicting rule wins, change permissions, or operate a live system.
  • Do not place unauthorized knowledge data or unnecessary personal, sensitive, or confidential data in an AI system.

The prompt

#CONTEXT:
Create a reviewable SOP draft from a supplied expert interview. Preserve provenance and uncertainty. The output is a draft for a named human owner, not a canonical-source update.

#INPUTS:
- Expert interview transcript with speaker labels: [INTERVIEW TRANSCRIPT]
- Named source owner and role: [SOURCE OWNER]
- Current system or location where this knowledge is stored: [SYSTEM OR LOCATION]
- Sensitivity label or classification: [SENSITIVITY]
- Supplied authority and decision rights: [AUTHORITY]
- Last-reviewed date, if documented: [LAST-REVIEWED DATE]
- Approved SOP structure or template: [SOP TEMPLATE]

#INSTRUCTIONS:
1. Extract only instructions, decisions, exceptions, prerequisites, and evidence present in the interview.
2. Cite the speaker and transcript passage for every operational statement. Never turn a suggestion or memory into a verified rule.
3. Build a reviewable SOP with purpose, scope, prerequisites, numbered steps, decision points, exceptions, stop conditions, evidence of completion, and escalation questions.
4. Label gaps Unknown. Label contradictory statements Unresolved conflict, cite both sources, and prepare the decision question for the named owner's review.
5. Separate the expert's documented authority from assumed organizational authority. Treat missing authority or review dates as Unknown.
6. Include a human review checklist and proposed verification exercise. Do not execute the process.
7. Maintain source attribution for every fact, rule, owner, authority claim, and date.
8. For each item, state its evidence status: Verified, Reported, Unknown, or Unresolved conflict.
9. Do not invent, infer, or fabricate rules, owners, authority, dates, facts, sources, or approvals.
10. Do not choose a winner or declare a source authoritative without documented supplied authority.
11. Use only authorized knowledge data in an approved AI workspace or approved AI vendor. Minimize the data; redact or omit unnecessary personal, sensitive, or confidential data. Preserve each sensitivity label or sensitivity classification and follow company retention and vendor policy.
12. Do not publish. Do not update a canonical source. Do not grant, revoke, or change access or permissions. Do not send material externally. Do not change a live system.
13. A named human owner must approve publication before action. A named human owner must approve every canonical-source update before action. A named human owner must approve every access or permission change before action. A named human owner must approve every external action or live system change before action. A named human owner must approve every policy exception before action.

#RESPONSE FORMAT:
## Source attribution and evidence status
| SOP claim | Evidence status | Speaker and transcript reference | Authority | Review note |
|---|---|---|---|---|

## Reviewable SOP
- Purpose and scope
- Preconditions
- Numbered procedure with decision points and stop conditions
- Exceptions and escalation path
- Evidence of completion

## Unknowns and Unresolved conflicts
| Item | Competing evidence | Named decision owner | Question to resolve |
|---|---|---|---|

## Human approval checklist

Input checklist

  • INTERVIEW TRANSCRIPT
  • SOURCE OWNER
  • SYSTEM OR LOCATION
  • SENSITIVITY
  • AUTHORITY
  • LAST-REVIEWED DATE
  • SOP TEMPLATE

Example input

Fictional example — INTERVIEW TRANSCRIPT: Maya explains how Northstar prepares a new-client workspace and says Finance, not Delivery, confirms billing status; she is unsure who approves an exception. SOURCE OWNER: Maya Chen, Delivery Lead. SYSTEM OR LOCATION: Operations handbook draft folder. SENSITIVITY: Internal; customer identifiers removed. AUTHORITY: Maya owns delivery steps but not finance policy. LAST-REVIEWED DATE: Unknown. SOP TEMPLATE: Purpose, scope, prerequisites, steps, exceptions, evidence, owner approval.

Expected output structure

  • A source-attributed, reviewable SOP with evidence states, numbered steps, explicit Unknowns, an Unresolved conflict register, and a human approval checklist.

Customize this prompt

  • Add transcript timestamps so reviewers can return to the exact statement quickly.
  • Use your approved SOP headings without removing provenance or uncertainty fields.

Guardrails

  • Maintain source attribution and an evidence status for every material claim; unsupported items remain Unknown and disagreements remain Unresolved conflict.
  • Do not invent, infer, or fabricate rules, owners, authority, dates, facts, sources, or approvals.
  • Do not choose a winner or declare a source authoritative without documented supplied authority.
  • A named human owner must approve publication before action.
  • A named human owner must approve every canonical-source update before action.
  • A named human owner must approve every access or permission change before action.
  • A named human owner must approve every external action or live system change before action.
  • A named human owner must approve every policy exception before action.
  • A named human owner and the qualified professional must approve every legal, tax, employment, or regulatory interpretation before action.
  • A named human owner must approve every payment or financial decision before action.
  • A named human owner must approve every deletion, closure, suspension, termination, revocation, deactivation, or archive action before action.
  • A named human owner must approve every promise or commitment before action.
  • Use only authorized knowledge data in an approved AI workspace or approved AI vendor; minimize data, redact or omit unnecessary personal, sensitive, or confidential data, preserve every sensitivity label or sensitivity classification, and follow company retention and vendor policy.
  • The assistant provides a reviewable draft only and must not publish, alter canonical sources, change access or permissions, act externally, or mutate a live system.

Related articles

NEXT STEP

Next step

Find out what your company's knowledge is worth.