Today
The product quality review needs a year of batches, test results, deviations, CAPAs, changes, complaints and stability data per product in one analysis. The data sits in LIMS, MES, the QMS and spreadsheets. QA pulls it together by hand for weeks, and little time is left for the actual assessment.
An example
At Rhenalis the agent gathers the 148 batches of the year for film-coated tablet FT-500: assay and dissolution from the LIMS as trends, 11 deviations, 4 change controls, 2 complaints and the 12-month stability results at 25 °C/60 % RH. It flags a dissolution trend towards the lower limit since an excipient supplier changed, and two CAPAs without an effectiveness check. The head of QA assesses both and signs the report.
Your data, wherever it lives
Today batches and yields sit in the MES, test results in the LIMS, deviations, CAPAs and change controls in the QMS, stability data in spreadsheets and earlier reports in the DMS. demi reads, understands and links them per product and year, with your rights. The AI calculates trends and drafts every section with its source, and QA spends its time on the assessment instead of collecting.
Where the data lives today
- MES
- Databases
- QM system
- ERP
- Office files
- DMS & archive
How demi solves it
Three layers in this order: connect the data first, then the rights, then the AI.
- 1 · ConnectData Layer
The case needs data from LIMS, MES, QMS, ERP and DMS. demi replaces nothing and links, per product and period, the batches with test results, deviations, CAPAs, change controls, complaints, stability data and suppliers. demi only writes back after approval.
- 2 · PermitPermission Layer
QA sees all products; a client's data only those who work for that client. Assessment and signature stay with QA and the Qualified Person, and every step is in the audit trail with person and time.
- 3 · ActIntelligence Layer
The agent calculates trends and figures deterministically from the records, compares them with the previous year and the specification limits and drafts each section to your PQR template. It names anomalies and missing data instead of smoothing them over; every figure points to its record.
How it runs
- 1
Set the PQR template and period per product
- 2
Bring together the period's batches, test results and deviations
- 3
Calculate trends against specification and previous year
- 4
Draft with anomalies and open points
- 5
QA assesses, adds and signs
What changes
Every PQR on schedule
QA assesses instead of collecting
Trends visible before the inspector asks
What does it bring you?
Check it with your own numbers
Every starting value is an assumption you can change. The calculation sits right below, step by step.
Your numbers
Result
+461 hfreed per year
Assumptions and calculation
demi gathers batches, deviations, changes and stability data per product; QA assesses and signs.
32 h → 22.4 h per case
How we calculate
4 PQR reports × 32 h × 30% × 1.0 (manual) × 12 months = 461 h
Value per year, before demi costs
Freed per year+461 h
Assumption: labour cost of €49.50 per hour in German manufacturing in 2025 (Destatis), plus workplace, IT and overheads.
= Value per year, before demi costs
+€29,965
gained per year
461 h ÷ 1,760 h per role = 0.3 full-time roles
An estimate, not a promise. The starting values are assumptions for an industrial company with 150 to 500 people.
Combine with more cases and get the packageQuestions about this use case
Which data goes into the PQR draft?
Everything your PQR template asks for that sits in your systems: batches and yields from the MES, test results from the LIMS, deviations, CAPAs, change controls and complaints from the QMS, plus stability data and supplier changes. How the data reaches demi, we clarify in a conversation.
Who assesses and signs the PQR?
QA, as your SOP defines. The agent delivers the draft with trends and open points. Assessment, conclusions and signature come from people, logged with person and time.
What if data is missing for a product?
The draft names the gap explicitly, for example a batch without a dissolution result or a stability time point still pending. The agent fills in nothing and calculates no trend on an incomplete basis.
Does a client see other clients' data?
No. In contract manufacturing demi keeps data separate per client. People working for a client see that client's products, and a report holds only data of the product it is written for.
Does this replace our PQR SOP?
No. Your SOP sets what the PQR contains and who assesses it. demi follows it and makes the collecting faster and traceable.
