Home · Business · Operations & Logistics · Quality control
As of 13 August 2026, AI can only partly create a quality management SOP.
This still needs a person who signs their name to it.
Can you do it?
15 minutesto a draft.
1 hourto something you’d act on.
Cost, all in£0
Skill neededchat-fluent
Who has to check ita colleague
What the alternative costsScribe is a purpose-built tool that turns a process you click through into a step-by-step guide automatically.
If this goes wrong: staff follow an incomplete or impractical control process, records are missed, and defective work may pass through the operation.
What to actually do
Hand it to a person
The route this page recommends
A person who owns the outcome does this end to end, worth it when the failure is dear.
Use a tool built for this
Second choiceDo it yourself
The distant thirdA chat interface, chat-fluent skill, and roughly 1 hour until you can act on the result.
How to actually do it
- Open the current work instruction, process map, inspection forms and non-conformance records for the process, and gather the applicable customer, internal and regulatory requirements.
- Ask the process owner and quality lead to confirm the process scope, responsible roles, acceptance criteria, inspection points, records, escalation rules and approval route.
- Remove unnecessary personal or commercially sensitive information, then paste the confirmed material into the prompt under Source material.
- Run the prompt in ChatGPT, Claude or Gemini and save the draft with a proposed document ID and version.
- Walk through the drafted numbered procedure at the work area with the person who performs the process, correcting every step, measurement, decision, role and record that does not match reality.
- Compare each acceptance criterion and record requirement against the current forms, equipment instructions and customer requirements, then resolve every [TO CONFIRM] item.
- Send the corrected SOP to the named process owner and quality lead for approval, publish the approved version in the controlled document location, and withdraw superseded copies.
Prompt
Create a controlled quality management SOP for [process name] at [organisation or site type]. Use only the information in the material below and do not invent requirements, measurements, responsibilities, equipment, records or compliance claims. If information is missing, mark it as [TO CONFIRM] and list the question needed to resolve it. Produce the SOP with these headings: 1. Document title, document ID, version, owner, approver and effective date 2. Purpose 3. Scope and exclusions 4. Definitions 5. Roles and responsibilities 6. Required equipment, materials and records 7. Procedure, written as numbered steps in the order performed 8. Quality acceptance criteria and how each criterion is measured 9. Inspection, testing and sampling points 10. Non-conformance, containment, corrective action and escalation 11. Required records, storage location and retention period, using [TO CONFIRM] where unknown 12. Training and competence requirements 13. Health, safety and environmental controls where supplied 14. Change control and review frequency 15. Revision history For every step, include the responsible role, input, action, output or acceptance decision, and record created. Separate facts from recommendations. Do not claim that the SOP meets ISO, regulatory or customer requirements unless the supplied material explicitly supports that claim. End with a table of open questions and a short approval checklist for the process owner and quality lead. Source material: [Paste the current work instruction, process map, inspection criteria, forms, responsibilities, known defects, escalation rules and applicable customer or internal requirements here.]
Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.
What it gets wrong
- AI cannot observe whether the written procedure matches what staff, equipment and materials actually do on site.
- AI cannot decide whether a proposed control is adequate for your product, process risks or customer requirements without competent human judgement.
- AI cannot validate measurements, sampling plans or equipment capability from a description alone.
- AI cannot approve the controlled document or transfer responsibility for a quality failure away from your organisation.
What caps this at PARTLY: judgement under ambiguity, verification cost and stakes of error.
How we scored this
Five axes, each scored nought to two by hand: ten means AI carries the task cleanly, and the thresholds that turn a total into YES, PARTLY or NO are published in the methodology. Each axis name links to its definition.
| Axis | Score (0–2) |
|---|---|
| Output | 2 |
| Inputs | 1 |
| Verification | 1 |
| Liability | 1 |
| Effort delta | 2 |
| Total | 7 / 10 |
The methodology and its thresholds are published in full.
FAQ
- Can ChatGPT write a quality management SOP?
- Yes, it can produce a useful first draft from your existing procedures, controls, records and responsibilities. A process owner and quality lead still need to test it against the real operation and approve the controlled version.
- Can AI make an SOP ISO compliant?
- Not by itself. AI can structure evidence you provide, but it cannot establish that your controls, records and implementation satisfy a particular ISO or customer requirement.
- What information does AI need to create a quality SOP?
- Give it the actual process steps, roles, equipment, materials, acceptance criteria, inspection points, records, non-conformance rules and approval route. Mark anything unknown instead of allowing the model to fill the gap.
- Who should approve an AI-written quality SOP?
- The process owner and the person responsible for quality in your organisation should check and approve it under your document-control process. They need to confirm that it is workable, accurate and supported by the required records and controls.
Nearby answers
- Can AI detect product defects from photos?PARTLY
- Can AI investigate the root cause of product defects?PARTLY
- Can AI verify my product labels?PARTLY
- Can AI check whether my supplier meets ISO 9001 requirements?PARTLY
- Can AI check printed packaging for errors?YES
- Can AI check my product's UKCA marking requirements?PARTLY
Assessed by gpt-5.6-luna (gpt-5.6-luna) on 2026-08-13, second-checked by an independent model. Wrong somewhere? Email [email protected] and it gets re-checked.
The newsletter
AI news, new answers and product picks, straight to your inbox.