Implementation guide, Platform
How to prepare for an ISO/IEC 17025 audit with AI
- Author
- PureField
- Published
- Updated
Preparing for an ISO/IEC 17025 audit with AI means having your LIMS records checked in the order an assessor would check them. The AI assistant in PureField LIMS runs a 20-check audit rehearsal (a mock audit), lists findings with clause, grade and record links, and builds a 30-day plan. The quality manager closes findings on the relevant screens and reruns the rehearsal.
Who is the audit rehearsal for?
It is for the quality manager of a laboratory preparing for an accreditation assessment, a scope extension or a surveillance visit. A month before the assessment, quality managers tend to walk through the same list by hand: is any instrument past its calibration date, is an internal QC violation still unassessed, has every CAPA had its effectiveness check, is any analyst's competence monitoring overdue.
Each of those checks lives on a different screen or in a different folder. The rehearsal reads them in one pass and sets out what is missing in the layout of an assessor's report.
What does the audit rehearsal check?
The rehearsal runs 20 fixed checks tied to ISO/IEC 17025 clauses, with the permissions of the signed-in person. The topics they cover:
| Topic checked | ISO/IEC 17025 clause |
|---|---|
| Impartiality declarations and current core documents | §4.1, §8.1, §8.2 |
| Mandatory training and competence monitoring | §6.2.5 |
| Environmental conditions | §6.3.3 |
| Calibration, maintenance and intermediate checks | §6.4 |
| Expired lots and supplier evaluations | §6.4.1, §6.6.2 |
| Method verification and validation | §7.2 |
| Formula verification | §7.11.2 |
| Internal QC and proficiency testing | §7.7 |
| Complaints and nonconforming work | §7.9, §7.10 |
| CAPA | §8.7 |
| Document control | §8.3 |
| Internal audit programme | §8.8 |
| Management review and quality objectives | §8.9, §8.2.2 |
| Accreditation record | Accreditation scope and expiry date |
The answer reads like an assessor's report: major findings first, then minor findings, then observations. Each finding names the clause, the count, up to five record links as objective evidence, and the screen where it is fixed.
How are finding grades and the readiness score calculated?
Grades and the readiness score come from fixed rules in PureField LIMS, not from the model. Because the rules are fixed, a rehearsal run twice on the same records gives the same grades; the model only explains the outcome.
- A major finding needs direct evidence of the gap in the records. For example: the latest approved validation study of an accredited method says "not fit for purpose", a proficiency testing area is past its participation cycle, a failed round has no CAPA linked to it, or the latest approved management review is more than 24 months old.
- A minor finding example: an instrument in service is past its calibration date. Only instruments with valid calibration can be added to an analysis, yet the overdue record is still visible at assessment.
- An observation example: an accredited method has no validation or verification record in the system. You show the paper record during the assessment.
- Readiness score = checks without a major or minor finding / checks performed × 100. Only major and minor findings lower the score.
- The score is given only when all 20 checks could run. If your permissions do not cover some sections, the answer shows how many checks ran (for example 7/20).
How do you use the 30-day preparation plan?
The plan ties each finding to a week, a responsible role and the screen where it is fixed. Major and minor findings come first; competence monitoring and a final rehearsal close the month. A sample plan looks like this:
| Week | Task | Owner | Screen |
|---|---|---|---|
| Week 1 | Schedule calibration for two overdue instruments | Quality manager | Equipment |
| Week 2 | Assess the Westgard violation on the pH method | Technical manager | Internal QC |
| Week 3 | Complete effectiveness checks on two CAPAs | Quality manager | CAPA |
| Week 4 | Finish overdue competence monitoring, run the rehearsal again | Laboratory manager | Staff |
In practice, this order works well:
- Run the rehearsal about a month before the assessment, from an account that can see every section; only then is a score calculated.
- Close major findings in the first week. If there is no approved management review, or it is more than 24 months old, hold one first.
- Open each finding from its record link and fix it on the relevant screen.
- Run the rehearsal again every week; findings you have closed drop off the list.
- In the last week, work out the likely assessment questions and get the evidence ready.
How is the management review agenda prepared from the same records?
The assistant collects the ISO/IEC 17025 §8.9.2 a–o inputs from the last 12 months of records, reads the last five management reviews, and suggests §8.9.3 outputs. If no review has been held before the assessment, this draft is where the meeting starts.
| §8.9.2 input | Records that feed the draft |
|---|---|
| b) Fulfilment of objectives | Quality objectives and actual values |
| d) Actions from previous reviews | Status of actions in earlier management review records |
| e) Internal audit outcomes | Internal audit findings with major, minor and observation counts |
| f) Corrective actions | Open and closed CAPAs, those awaiting an effectiveness check |
| i–j) Feedback and complaints | Satisfaction scores and complaint records |
| l) Adequacy of resources | Instruments past calibration, pending staff authorisations |
| n) Validity of results | Westgard violations and proficiency testing scores |
The draft is an input to the meeting; the review record, freezing the agenda and approval happen on the management review screen.
Which questions will the assessor ask?
The assistant derives likely questions from two sources: the rehearsal's findings and the "frequently checked in assessments" list on its TÜRKAK reference card. TÜRKAK is Türkiye's national accreditation body. For each question it names the evidence you will show and the screen it is on.
If the rehearsal found an unassessed Westgard violation on the pH method, for instance, it notes that the assessor is likely to ask how the violation was assessed and why that day's results were approved, and that the evidence is the violation record on the internal QC screen.
How should you read the rehearsal result?
The rehearsal is a pre-assessment; the accreditation decision belongs to the assessor. Every rehearsal answer therefore carries a fixed note, independent of what the model wrote: "A PureField LIMS pre-assessment — not a decision of the TÜRKAK assessor; check the records on their screens."
Use the score to see where preparation is falling short. As it rises, confirm on each record's own screen that the findings are really closed.
How do you start a rehearsal in PureField LIMS?
In PureField LIMS you start a rehearsal in one of two ways: the AI audit rehearsal button on the 17025 compliance map, or by asking the assistant "Are we ready for the TÜRKAK assessment?". The map lists ISO/IEC 17025 clauses in order, each with the screens that support it and the evidence you have prepared.
For the management review agenda, ask the assistant to prepare meeting notes for the management review. You can copy the answer into the minutes or download it as Markdown.
To see the rehearsal on records close to your sample types, request a PureField LIMS account.