Sample finding report

This is what you receive on the afternoon of the second day of a review. The data is synthetic: the fictional company Demo Pharma, 25 documents, reviewed in our published test 2.

Sample · not a client documentThe 16 possible findings come from the output of the fixed rules and the local AI on 2 November 2026. For this sample the output was translated from Estonian and shortened, and two duplicates were removed. The model's own confidence rating is left out, because it is not a measured score. The suggested actions, the order of work and the clause numbers in findings 1, 15 and 16 were written by the reviewer. Download the document set and the answer key (ZIP).

Finding reportGMP document review · Eesti Farmaatsia Grupp OÜ

Client: Demo Pharma OÜ (fictional)
Site: Demo Pharma OÜ manufacturing site, Estonia (fictional)
Reviewed by: Veiko Saluste, Eesti Farmaatsia Grupp OÜ
Date of the review: 2 November 2026
Report: Version 1.0, sample
Documents reviewed: 25, as in the list agreed in advance: 1 quality manual, 21 SOPs, 2 specifications and 1 set of 5 forms. Language: Estonian.
File list: DP-FL-001, version 1: the 25 reviewed files with their versions. In this sample it is the content of the ZIP file.
Tools: Fixed rules and a local AI model on EFG's own machine, offline. Rule check under one second, AI part 2.2 minutes.

1. Summary

Checked: 25 documents from the agreed list: references between documents, content that differs between documents, and the EU GMP clauses listed in section 2. Records and practice were not checked.
For your QA to assess first: findings 1, 2, 3, 4, 6, 15 and 16. They concern batch certification, a release limit, aseptic processing, environmental monitoring, sterilisation, complaints and recalls. Section 4 gives the suggested order.
Known miss: this check did not find one planted error, a Grade A action limit that differs between two procedures. See section 5.
Result: 16 possible findings, 8 from fixed rules and 8 from local AI, after 2 duplicates were removed. Each one is a possible finding until your QA has checked it against the source.

2. Scope

Checked

Document codes, titles, versions and annexes referred to in other documents (fixed rules).
Whether the processes named in EU GMP Chapter 8 have a document (fixed rules).
Limits, deadlines, roles and equipment that differ between documents (local AI).
The procedures against the EU GMP clauses supplied as text: Annex 1 §9.30 and §9.32–9.38, Chapter 4 §4.11–4.12, Chapter 8 §8.5 and Chapter 9 §9.1–9.2, and one sentence stating the principle of Annex 16 that a batch is certified by a Qualified Person (local AI).

Not checked

Records, batch documentation and whether the procedures are followed in practice.
Signatures and approval dates: the test documents are text files without signature pages.
EU GMP requirements other than the clauses listed.
Whether methods, limits and validation are technically right.

3. Findings

A. Content that differs between documents or from EU GMP

Finding 1RoleLocal AI

DocumentsSOP-QA-009 Batch release, version 3, §3 Possible findingThe procedure makes the QA manager responsible for certifying and releasing a batch. Under EU GMP a batch is certified by a Qualified Person (QP). Quoted textThe QA manager certifies and releases the batch. BasisEU GMP Annex 16, given to the model as one sentence: a batch is certified by a QP. Clause added by the reviewer: §1.1. Suggested actionFirst confirm from the batch register who has certified batches since version 3 took effect on 01.10.2025, and assess the impact if it was not the QP. Then correct §3 so that the QP certifies the batch; QM-001 §3 already says so. QA decision☐ Confirmed☐ RejectedReason Follow-up: ☐ none ☐ deviation ☐ CAPA ☐ change controlReference Owner, due date Name, date

Finding 2LimitLocal AI

DocumentsSOP-QC-002 Finished product testing, version 4, §4.2 · SPEC-FP-001 Finished product specification, version 3, §2 Possible findingThe acceptance criterion for assay is 90.0–110.0 % in the testing procedure and 95.0–105.0 % in the specification. The two documents should give the same limit. Quoted textSOP-QC-002: Assay: 90.0–110.0 % of the declared content. SPEC-FP-001: Assay: 95.0–105.0 % of the declared content. BasisBetween documents. Suggested actionFirst confirm which limit is the approved one; the registered specification and the marketing authorisation decide. Review the assay results of batches tested since 01.08.2025 against that limit and assess any result outside it. Then correct the document that is wrong, through change control. QA decision☐ Confirmed☐ RejectedReason Follow-up: ☐ none ☐ deviation ☐ CAPA ☐ change controlReference Owner, due date Name, date

Finding 3RegulationLocal AI

DocumentsSOP-PR-005 Aseptic process simulation, version 3, §4.2 Possible findingThe procedure sets the periodic aseptic process simulation (APS) at once a year. Annex 1 (2022) says APS should normally be repeated twice a year, approximately every six months. Quoted textPeriodic repeat: one APS once a year per filling line. BasisEU GMP Annex 1 (2022) §9.38, supplied to the model as text. Suggested actionFirst check when APS was actually performed for each aseptic process, filling line and shift. If an interval was longer than about six months, assess the impact and decide on interim measures. Then set the frequency in the procedure through change control. Any other frequency is an exception that your QA has to justify and approve. QA decision☐ Confirmed☐ RejectedReason Follow-up: ☐ none ☐ deviation ☐ CAPA ☐ change controlReference Owner, due date Name, date

Finding 4Outdated referenceLocal AI

DocumentsSOP-PR-003 Environmental monitoring, version 3, §1 Possible findingThe procedure states that it is based on the 2008 version of Annex 1. The 2022 version has been in operation since 25 August 2023. Quoted textThe procedure is based on EudraLex Volume 4, Annex 1, version 2008. BasisEU GMP Annex 1 (2022), in operation since 25 August 2023. Suggested actionFirst confirm which limits and monitoring are applied in practice and compare them with Annex 1 (2022); see also section 5. Then review the whole procedure against Annex 1 (2022), not only the reference, through change control. QA decision☐ Confirmed☐ RejectedReason Follow-up: ☐ none ☐ deviation ☐ CAPA ☐ change controlReference Owner, due date Name, date

Finding 5DeadlineLocal AI

DocumentsQM-001 Quality manual, version 4, §5 · SOP-QA-003 Deviation handling, version 5, §4.5 Possible findingThe quality manual gives 20 working days for investigating and closing a deviation; the procedure gives 30 calendar days. Quoted textQM-001: A deviation is investigated and closed within 20 working days of detection. SOP-QA-003: A deviation is investigated and closed within 30 calendar days of detection. BasisBetween documents. Suggested actionDecide which deadline applies and correct the other document. QA decision☐ Confirmed☐ RejectedReason Follow-up: ☐ none ☐ deviation ☐ CAPA ☐ change controlReference Owner, due date Name, date

Finding 6EquipmentLocal AI

DocumentsSOP-EN-002 Autoclave sterilisation, version 4, §2 · SOP-EN-001 Equipment qualification, version 3, §5 Possible findingThe scope of the sterilisation procedure is autoclaves AK-01 and AK-02. According to the equipment list AK-02 was retired on 30.11.2025, and AK-03, in use since 01.12.2025, is not covered by the procedure. Quoted textSOP-EN-002 §2: Autoclaves AK-01 and AK-02. SOP-EN-001 §5: AK-02: retired from use 30.11.2025. AK-03: in use since 01.12.2025. BasisBetween documents. Suggested actionFirst confirm that AK-03 is qualified, that its loading patterns are validated and which instruction operators follow today. If any of this is missing, assess the loads sterilised since 01.12.2025. Then update the scope of SOP-EN-002 and check that change CC-2025-041 listed this procedure among the affected documents. QA decision☐ Confirmed☐ RejectedReason Follow-up: ☐ none ☐ deviation ☐ CAPA ☐ change controlReference Owner, due date Name, date

Finding 7Circular referenceLocal AI

DocumentsSOP-QA-001 Document management, version 3, §4.2 · SOP-QA-006 Change management, version 4, §4.3 Possible findingEach document points to the other for the rules on numbering and labelling documents. Quoted textSOP-QA-001 §4.2: The rules for numbering and labelling documents are described in SOP-QA-006 Change management. SOP-QA-006 §4.3: … the rules for numbering and labelling documents, which are described in SOP-QA-001 Document management. BasisBetween documents. Suggested actionDescribe the rules in one of the two documents and point the other one to it. QA decision☐ Confirmed☐ RejectedReason Follow-up: ☐ none ☐ deviation ☐ CAPA ☐ change controlReference Owner, due date Name, date

Finding 8Process linkLocal AIreason to be checked

DocumentsSOP-QA-004 CAPA, version 3, §4.3 · SOP-QA-007 Self-inspection, version 2, §4.1 Possible findingThe CAPA procedure sends the effectiveness check to the next self-inspection. The self-inspection plan covers each area at least once a year, so the check could take place up to a year after the action. Quoted textSOP-QA-004 §4.3: The effectiveness check is carried out during the next self-inspection in accordance with SOP-QA-007 Self-inspection. SOP-QA-007 §4.1: The audit plan covers all GMP areas at least once a year. BasisBetween documents. Reviewer's noteThe model named the right pair of documents but not the main gap: SOP-QA-007 does not describe a CAPA effectiveness check at all. In our test this finding counts as found in part. Suggested actionDefine where, when and by whom CAPA effectiveness is checked, and describe it in one procedure. QA decision☐ Confirmed☐ RejectedReason Follow-up: ☐ none ☐ deviation ☐ CAPA ☐ change controlReference Owner, due date Name, date

B. References to documents, versions and annexes

Finding 9Wrong versionFixed rule

DocumentsSOP-PR-004 Aseptic filling, version 5, §4.1 Possible findingThe procedure refers to version 2 of SOP-PR-005 Aseptic process simulation. Version 3 is in force. BasisDocument register: SOP-PR-005 version 3, effective 01.03.2026. Suggested actionCorrect the reference, or refer to the procedure without a version number. QA decision☐ Confirmed☐ RejectedReason Follow-up: ☐ none ☐ deviation ☐ CAPA ☐ change controlReference Owner, due date Name, date

Finding 10Wrong titleFixed rule

DocumentsQM-001 Quality manual, version 4, §4 Possible findingThe process table calls SOP-QA-010 “Complaint handling”. Its actual title is “Supplier qualification”. BasisDocument register. Suggested actionCorrect the table. See also findings 13 and 15. QA decision☐ Confirmed☐ RejectedReason Follow-up: ☐ none ☐ deviation ☐ CAPA ☐ change controlReference Owner, due date Name, date

Finding 11Wrong titleFixed rule

DocumentsSOP-QA-008 Product quality review, version 3, §4.2 and §5 Possible findingSOP-QC-003 is called “Stability studies”. Its actual title is “Investigation of OOS results”. The set contains no procedure for the stability programme. BasisDocument register. Suggested actionCorrect the reference and decide which document describes the stability programme. QA decision☐ Confirmed☐ RejectedReason Follow-up: ☐ none ☐ deviation ☐ CAPA ☐ change controlReference Owner, due date Name, date

Finding 12Missing documentFixed rule

DocumentsSOP-QA-002 Training, version 2, §4.5 and §5 Possible findingThe procedure refers to form FRM-QA-006 “Assessment of training effectiveness”. The set contains no such form; the forms end at FRM-QA-005. BasisDocument register. Suggested actionCreate the form or correct the reference. QA decision☐ Confirmed☐ RejectedReason Follow-up: ☐ none ☐ deviation ☐ CAPA ☐ change controlReference Owner, due date Name, date

Finding 13Missing documentFixed rule

DocumentsSOP-QA-003 Deviation handling, version 5, §4.6 and §5 Possible findingThe procedure refers to SOP-QA-011 “Complaint handling”. The set contains no such document. BasisDocument register. Suggested actionSee finding 15. QA decision☐ Confirmed☐ RejectedReason Follow-up: ☐ none ☐ deviation ☐ CAPA ☐ change controlReference Owner, due date Name, date

Finding 14Missing annexFixed rule

DocumentsSOP-QC-001 Receipt and sampling of starting materials, version 3, §4.3 Possible findingThe procedure refers to the sampling plan in Annex 2. The document has only Annex 1. BasisThe document itself. Suggested actionAdd the sampling plan or correct the reference. QA decision☐ Confirmed☐ RejectedReason Follow-up: ☐ none ☐ deviation ☐ CAPA ☐ change controlReference Owner, due date Name, date

C. Procedures that appear to be missing

Finding 15Missing procedureFixed rule

DocumentsWhole set Possible findingThe reviewed set contains no document on complaints. The check cannot tell whether the procedure is part of another document or was left out of the export. BasisEU GMP Chapter 8 §8.5 expects written procedures for the actions to be taken when a complaint is received. Suggested actionConfirm whether the procedure exists in another document or outside the reviewed set. If it does not, decide how a complaint received today is handled, and write the procedure. QA decision☐ Confirmed☐ RejectedReason Follow-up: ☐ none ☐ deviation ☐ CAPA ☐ change controlReference Owner, due date Name, date

Finding 16Missing procedureFixed rule

DocumentsWhole set Possible findingThe reviewed set contains no document on recalls. The check cannot tell whether the procedure is part of another document or was left out of the export. BasisEU GMP Chapter 8 §8.20 expects established written procedures for recalls. Suggested actionConfirm whether the procedure exists in another document or outside the reviewed set. If it does not, name now who decides on a recall and carries it out, and write the procedure. QA decision☐ Confirmed☐ RejectedReason Follow-up: ☐ none ☐ deviation ☐ CAPA ☐ change controlReference Owner, due date Name, date

4. Suggested order of work

This report covers documents only. Whether a finding has affected product can be judged only by your QA, from your records. For that reason the first step is an assessment, and the correction of the document comes second.

First: assess the impact and decide on interim measures

Finding 1: who has certified batches since version 3 took effect.
Finding 2: which assay limit is the approved one, and whether any result since 01.08.2025 falls outside it.
Finding 3: when APS was actually performed for each aseptic process, filling line and shift.
Finding 4 and the missed limit in section 5: which environmental monitoring limits are applied in practice.
Finding 6: whether AK-03 is qualified, whether its loading patterns are validated, and which instruction operators follow today.
Findings 15 and 16: whether procedures for complaints and recalls exist outside the reviewed set. If there is none for recalls, who decides on a recall and carries it out today.

Then: permanent correction through change control

Findings 1 to 4, 6, 8, 15 and 16, each with an owner and a date set by your QA after the assessment.

At the next revision of the document

Findings 5, 7 and 9 to 14.

This order is the reviewer's suggestion. Your QA sets the priorities and the dates.

5. What this report missed

Because this document set is synthetic, we know what was planted in it: fifteen errors and one gap. The report contains thirteen of the errors in full and one in part (finding 8). One was missed: the Grade A action limit is “1 CFU” in SOP-PR-003 §4.3 and “no growth” in SOP-QC-004 §4.2. Finding 16 was not planted: the set as we wrote it has no recall procedure. On your own documents nobody knows in advance what is there to find, so a report like this shortens a review. It does not replace one.

6. Handover

Handed over by (EFG)Name Date, signature

Received by (client QA)Name, role Date, signature

The signature confirms receipt of the report. Your QA decides on each finding afterwards. Comments on the report reach us within ten working days of receipt. Unless we have agreed otherwise in writing, EFG keeps no copy of this report.

Discuss your review → How we check → How this test was scored →