GMP training records for inspection requests, as a cited evidence pack in minutes
Every inspector's training question answered with a cited evidence pack in minutes, each gap explained to QA first.
See one case, screen by screen ↓The lead inspector asks a simple question: show me that everyone who touched this batch was trained on the SOP versions in effect when they did the work. To answer it, someone reads the executed batch record step by step, writes down who performed and who verified each one, finds the procedure revision that was effective on that date, and then pulls each person’s transcript from the learning system and their media fill and gowning status from the qualification register.
That is three or four systems and a few hours for one request — while the inspector waits in the front room and more requests arrive. And the hard part is not the green rows. It is the operator who completed the new revision two days after the work, or the media fill that sits inside the grace window. If QA first sees that gap when the inspector does, the explanation already on file — the deviation, the leave record, the booking — looks like an afterthought.
Estimated split for one batch request, in hours, by hand and with the solution.
Six specialist agents read the request, the batch record, each procedure revision and every training record, then build the pack; the QA lead decides each gap and signs the release.
Day 2 of a four-day agency inspection at Plant 2 — Westport, where Veltrimab 150 mg/mL vials are filled on Fill Line 2. At 10:31 the lead inspector asks about batch VB-26-0917. Here is what happens next, screen by screen, in the working solution.
Grace opens the back room. Six training requests so far today, two answered and four open; a median of 18 minutes to a pack against a 30-minute target; three gaps found before the inspector. On the right, who is in conference room B — Ruth Calder, lead inspector, and Ben Ashworth — and the 11:00 training records review, where R-14 and R-17 are expected.
“What did the inspector ask? Type or paste the request as the scribe wrote it.”
Sam logs each request verbatim, with who asked and when. The request log shows all seven from this inspection on their clocks: R-14 asked, R-15 with the agents, R-16 waiting for QA on two people, and R-11, R-09 and R-06 already with the inspector in 21, 13 and 18 minutes.
“Show me that everyone who touched batch VB-26-0917 was trained on the SOP versions in effect when they did the work.” The request opens with the inspector’s words on top and the scribe’s log beside it, highlighted. The clock is already running against the 30-minute target.
The agents read only; the batch record, learning and document systems stay the records of truth.
The request reader sets the scope: batch VB-26-0917, Sep 16–18, everyone who performed or verified a step. The batch record reader reads 24 pages — 9 steps, 8 people, names joined by employee number. The SOP version resolver finds the revision in effect at each step and notes that SOP-MFG-0262 rev 7 took effect 9 days before. The checker runs 18 training checks and 8 qualification checks.
Batch steps, people and procedure revisions on one trace: 8 people, 9 steps, 24 of 26 checks passed, 2 to explain. Two lines stand out — Lucas Moreau to SOP-MFG-0262 rev 7, “trained Sep 19”, and Ciarán Doyle to his media fill, “grace day 12”. Behind the trace, the evidence matrix lists all 26 checks, each with the batch record page and the transcript or register entry it rests on.
Lucas Moreau capped vials at 14:20 on Sep 17. Rev 7 of stoppering and capping took effect Sep 8; he completed it Sep 19 — assigned Sep 1, due Oct 8. The gap explainer finds that Lena Ortiz raised DEV-2026-0398 for this event on Sep 22: the torque checks were still done every 30 minutes as rev 7 requires, verified by Aoife Byrne, trained on rev 7 on Sep 4. His transcript opens beside it.
A gap is never left out of a pack — QA decides how it is presented.
Grace chooses “Disclose with DEV-2026-0398”. Next, Ciarán Doyle: he entered Grade A on Sep 16, his last media fill was Mar 4, so the 6-month date was Sep 4 — he was on leave for the Aug 26 run and is booked for Oct 14. SOP-QA-0150 §6.2 allows a 30-day grace window, so Sep 16 is day 12. She confirms it inside the window. Each decision is recorded in the audit trail.
The release names what she is signing: R-14, 26 checks, two gaps with her decisions. She signs with her name and role, the meaning — “QA approved — release to the inspector” — and her password, and the pack becomes read-only.
The response letter answers the question first: 7 of 8 people completed training on the revision in effect before the work; Lucas Moreau did not — with the disclosure citing DEV-2026-0398 and CAPA-2026-0151 — and Ciarán Doyle worked inside the media-fill grace window. 24 of 26 checks were met. Signed by Grace Mensah and handed to Dana Okafor for the front room, read-only.
“The learning, batch record and document systems remain the records of truth; this pack is a read-only extract.”
When the inspector follows up, the look-up answers on the spot. Lucas Moreau on Sep 17, 2026: trained on gowning rev 8 and aseptic filling rev 11; stoppering and capping rev 7 “Not yet · due Oct 8”. Media fill valid to Feb 26, 2027; worked that day on aseptic filling at 06:30 and capping at 14:20. Slide the date and the picture changes.
The same request, seen by the four people who carry it — from the QA lead who signs to the host who hands it over.
Six specialist agents read the request, the batch record, each procedure revision and every training record, then build the pack; the QA lead decides each gap and signs the release.
Reads the inspector’s request as the scribe logged it and sets the scope — batch, equipment, procedure or person — with the dates and the people to check. By default, everyone who performed or verified a step.
Reads the executed batch record or logbook and lists every step with who performed and who verified it, the time, and the procedure the step names — keeping the page and entry for every field.
Finds the revision of each procedure that was in effect at the moment of each step, and flags revisions that took effect within 30 days before the work.
Checks each person completed the revision in effect before the step, and their media fill, gowning or Knapp qualification at that moment, using the rules in Settings.
For every gap, searches deviations, CAPA, leave and access records for the explanation already on file, and proposes how QA should present it: disclose with the record, confirm inside a window, or hold and investigate.
Builds the evidence pack — index, the request verbatim, a summary table, disclosures, then every source extract stamped with its print time — and a short response letter that answers the question in its first sentence.
Decides each gap and signs. The agents propose; a named person decides.
The back room can ask in plain words — or tell it what to change. Every answer opens its source record.
Was everyone on batch VB-26-0917 trained on the SOPs in effect?
Not quite — 1 gap and 1 check. Batch VB-26-0917 was made by 8 people across 9 steps, Sep 16–18; 24 of 26 checks pass. Lucas Moreau’s is already covered by DEV-2026-0398; his transcript and the batch record signature are in the pack.
Why is Lucas Moreau a gap?
SOP-MFG-0262 rev 7 took effect on Sep 8 and changed the cap torque check from every 60 to every 30 minutes. Lucas capped batch VB-26-0917 at 14:20 on Sep 17 and completed rev 7 on Sep 19 — it was assigned Sep 1 with a due date of Oct 8. The torque checks were still done every 30 minutes, verified by Aoife Byrne, who was trained on rev 7.
Who was not trained on gowning rev 9 before Oct 1?
Two of 38 people in the aseptic curriculum. Emma Kelly — not trained; on leave since Sep 20 with her Grade A/B badge suspended, no entries since. Jonas Weber — trained Oct 2 at 08:10; no Grade A/B entry before then. The other 36 finished between Sep 14 and Sep 30.
Add a rule: flag training completed within 24 hours of the work
Done. New check: flag training completed within 24 hours before the person performs or verifies a step — it shows as a note for QA, not a gap. Re-run on the open packs: nothing is flagged today. The rule is live in Settings and recorded in the audit trail.
13 screens from the working solution, on its sample data. Pick one to see it large.
The checks follow your procedures. Each rule is a setting, versioned and approved.
| Setting | Default | Choose from |
|---|---|---|
| Response target, from the scribe’s log to a QA-approved pack | 30 min | 15 to 60 min, in steps of 5 |
| Media fill grace window after the 6-month date | 30 days | 0 · 15 · 30 · 45 days |
| Include verifiers | On | On · Off |
| Match names by employee number | On | On · Off |
| Add a rule | Your rules | Describe it in one sentence, for your approval |
| QA check — decides how each gap is presented | Grace Mensah | Grace Mensah · Priya Raman |
| Release to the inspector — electronic signature | Grace Mensah | Grace Mensah · Omar Haddad |
| Second approval for a gap disclosed without a deviation | Omar Haddad | Omar Haddad · Grace Mensah |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample inspection data · “target” = the design goal, measured in the live solution · “estimated” = our estimate (3–4 h by hand for one batch request) · “typical” = published figures (FDA warning letters citing training gaps under 21 CFR 211.25(a), 2025–2026). People, companies and products named on this page are fictional — characters and sample data in the working solution.
It is the answer to an inspector’s training question: every person who performed or verified a step, the SOP revision in effect at that moment, and the date they completed it, with the source records attached. This solution builds it from the batch record, the learning system, the document system and the qualification register, and adds a short response letter.
The SOP version resolver takes the procedure named in the batch record for each step and returns the latest revision whose effective date is on or before the step — the effective date, never the approval date. It flags revisions that took effect within 30 days before the work and cites the version history row.
Yes. For Grade A/B steps it checks media fill every 6 months plus the grace window in Settings, and gowning yearly; for visual inspection it checks the yearly Knapp test. Each result cites the register entry.
The gap stays in the pack. The gap explainer searches deviations, CAPA, leave and access records for the explanation already on file and proposes how to present it. QA decides — disclose with the record, confirm inside a window, or hold and investigate — before anything goes to the inspector.
No. The agents read the source systems and never change a training, batch or quality record. The released pack is a read-only extract, and every line opens the source entry it came from.
Yes. The agents propose; the QA lead decides each gap and releases the pack with an electronic signature that records name, meaning and time. A gap disclosed without a deviation needs a second approval from the head of site quality.
Yes. The request reader sets the scope from the inspector’s words — a batch, a piece of equipment such as lyophiliser LY-02, a procedure revision, a process or a person — and the same checks run on it. The look-up shows anyone’s training status on any date.
The Agentic Solution Engine builds and deploys it from your requirements and documents — your procedures, your check rules and a sample of past batch records and training history — and it goes live once every quality gate has passed. We will walk you through it on one of your own batches first.
Every SOP change retrains only the people it touches, all trained before it takes effect.
1-min filmHR AnswersEvery employee gets the policy answer for their own country in seconds, with the clause quoted.
1-min filmPay TransparencyEvery pay-information request answered in days, every pay gap explained with evidence.
1-min filmWe’ll run the Training Evidence Pack on one of your own batches and its training records.