Quality History and Evidence

Quality history is the evidence corpus behind the Audit Prep assistant. It is built automatically from uploaded client files and linked quality records. Evidence is stored as searchable content scoped strictly to each individual client.

Table of contents

Before you start

Before you start

The client must have at least one file uploaded in the Files tab.
The EnableDocumentSignals feature flag must be enabled.
For AI-assisted analysis, the EnableQualityHistory flag must also be enabled.

Evidence types and sources

The following record types contribute to a client's quality history corpus:

Uploaded files: PDFs, Word documents, and spreadsheets uploaded via the Files tab.
Discovery documents: structured quality records linked to the client during onboarding.
CAPA records: corrective and preventive action reports.
NCR records: non-conformance reports.
5S submissions: 5S audit inspection findings linked to this client.
SPC inspection records: statistical process control measurements.

Uploading and verifying evidence

Open the client dashboard and select the Files tab. -> Uploaded files are listed with name, date, and category.
Upload one or more quality documents using the upload control. -> Each file is processed and ingested into the quality-history pipeline. Ingestion may take up to 60 seconds per document depending on file size.
Return to the Audit Prep tab or the Impact tab to see evidence reflected. -> Newly ingested files appear as citations in gap analysis and impact signals.
Use the AI Ask feature (available in Audit Prep clause detail and Impact tab) to ask a natural-language question about this client's quality history. -> The answer is grounded: it cites the specific document and page it drew from.
Review citations carefully. -> Each citation shows the file title and page number. If a citation references an unexpected document, verify that the correct files are uploaded for this client.

Common issues

Ask returns no answer found

The question may be too broad or the relevant documents have not been uploaded yet. Try a more specific question scoped to one clause, or upload additional files and try again.

Citations are empty after an analysis

Evidence was scored but the source document contained no extractable page-level citation data. This can happen with scanned images that lack OCR text. Use text-based PDFs for best citation accuracy.

Evidence from a previous client appears

Evidence is strictly scoped per client and this should not happen. If you suspect cross-client data, report it to your administrator immediately.

File upload fails

Ensure the file is a supported format (PDF, DOCX, XLSX) and under the maximum file size limit. Try uploading one file at a time if batch uploads fail.

Tips

Use structured records for best clause matching

Upload structured records such as CAPA and NCR reports rather than raw meeting notes. Structured records produce more accurate clause-level citation matches.

Name files descriptively

Name files descriptively (e.g., "2025-CAPA-032-Gauge-Calibration.pdf") so citations in gap analysis are easy to trace back to their source during client conversations.

Upload calibration and MSA records

Calibration records and MSA studies improve coverage of measurement system and control plan clauses, which are commonly scored as gaps when files are missing.

What to do next

Run audit prep

After uploading evidence, run the Audit Prep assistant to see updated clause scores. More evidence = more covered clauses.

Fill citation gaps

After identifying citation gaps in Audit Prep, return here to upload the missing records for the clauses that still show empty citations.