GCP inspection readiness where every storyboard sentence is backed by the record
Every likely inspection topic backed line by line by the record.
See one case, screen by screen ↓Software that gets a sponsor ready for an FDA or MHRA GCP inspection: the topics an inspector will probe, a factual storyboard for each, the gaps in the record, and rehearsed answers for the front room. GCP Inspection Readiness adds six agents that read the trial record in place, draft storyboards with a citation on every sentence, check each sentence against its source and play the inspector — the QA lead signs each storyboard.
How to choose AI solutions for life sciences →
Updated
When FDA or MHRA announces a GCP inspection, clinical QA has weeks to make every topic an inspector may probe — out-of-window visits, consent on a superseded form, data changes after soft lock, oversight of the CRO — into one factual story, and to prepare the people who will answer in the front room. Today each storyboard is assembled by hand from the eTMF, the CTMS, the EDC and the quality system, in slides and spreadsheets.
The risk sits in the sentences that sound right but are not in the record: “the new coordinator was trained before she booked her first visit”, “every change was approved before hard lock”. Inspectors test exactly those claims against the TMF. Documents filed after the notice show in the eTMF audit trail, and a back room that takes half an hour to find a visit report reads as a disorganised TMF.
Estimated split for one storyboard on a sponsor inspection, by hand and with the solution.
Six specialist agents index the record, find topics, draft, check, rehearse and fetch documents; the QA lead approves every storyboard.
Northwind Bio’s VLT-302 — Veltrimab in plaque psoriasis, licence application under review. The FDA notice arrived on Oct 5 for a sponsor inspection in Cambridge, MA, Mon Oct 26 – Thu Oct 29. Here is Oct 7, screen by screen, in the working solution.
Dana opens on the countdown: 19 days to the opening meeting, readiness 84/100 and +8 this week, 4 of 12 storyboards inspection-ready, 274 of 291 sentences backed, 14 open gaps. The readiness radar scores nine inspection areas; Protocol compliance at 72 and Data handling at 76 sit below 80. The storyboard wall shows every frame of every story, coloured by whether the record supports it.
The inspector is likely to ask: “Three Week 16 visits at Site 114 were out of window. How do you know your corrective action worked?” — Dr. Maya Chen · not rehearsed.
Clicking Protocol compliance opens SB-01, Week 16 visits out of window at Site 114, at frame 6 — What we did. CAPA-2026-031 retrained every member of the Site 114 team on 20 March 2026; that sentence is cited. The next one — the new coordinator, who started on 2 March, was trained on the visit schedule before she booked her first visit — is underlined red. Overnight the Claim Checker re-read all 31 claims: 30 supported, 1 not in the record.
“The record cannot support this yet.” Actions in the CAPA: “Train new coordinator before booking visits — No record.”
The citation opens the Site 114 staff training log, filed in TMF 05.02.08 and read in place — nothing copied. K. Brooks has Protocol v4.0 on 3 March and EDC user training on 4 March; her first visit-schedule training is the 20 March retraining. She booked the Week 16 visit for 114-0019 on 9 March. Gap G-01 is open for Omar Haddad, due Oct 14: find an earlier training record, or reword the claim.
“No visit-schedule training is recorded for K. Brooks before 20 Mar 2026.”
The week plan lays out 12 sessions over four days with 6 people in the front room; Wednesday’s 09:00 protocol compliance session, Dr. Maya Chen with SB-01, is flagged. Beside it, what the record cannot back: six statements struck through with the reason. 38 of the 214 changes after soft lock have no approval entry in the change log; the reopening memo is a draft, v0.3, not signed; Halvorsen’s oversight minutes for July and August 2026 are not in the TMF.
“Statements in a storyboard that no document proves. An inspector would find each one.”
One click shows the answer from the record: the whole Site 114 team retrained on 20 March under CAPA-2026-031; since then all 9 Week 16 visits at the site were inside the window, a median of 1 day from the target date; the CAPA closed as effective on 30 June; Week 16 windows are checked at every on-site visit. Every sentence carries its citation, and the coaching for Dr. Chen is underneath.
Coaching: “Do not say the new coordinator was trained before 9 March — say she was retrained with the whole team on 20 March.”
Dr. Chen answers as she would in the front room. The check reads her four sentences against the record: two backed; the claim that the new coordinator was trained before she booked any visit is not in the record; “our monitors check visit windows at every visit” is only partly backed — monitoring plan v3 checks it at on-site visits, not remote ones. In the back room, four document requests from the Oct 5 mock went out at a median of 0:41.
Front-room habits: “If you do not know, say you will check. Never guess.”
Fourteen gaps are open across 8 storyboards, 4 of them high severity and 5 due by Wed Oct 14, each with what is needed, the frame it blocks and an owner. The late-filing watch counts 37 documents filed Oct 5–7 — 29 of them dated before July — and only 11 with a note-to-file.
“Filed after the inspection notice — inspectors read the audit trail.”
Dana clicks Draft the missing notes. The solution reads the 26 late filings without a note — document date, filed date, who filed, why it was late — and drafts one note per document with a factual reason from the filing history. The notes go to Omar Haddad, the TMF manager, for signature.
“Notes are filed only after he signs.”
SB-03, Emergency unblinding at Site 209, has 21 sentences supported, 1 partly and none outside the record. Dana chooses the meaning of her signature — “Approved as factual and complete” or “Approved with the noted gaps” — and signs with her password. The partly supported sentence stays an open gap after she signs. Approving with a statement not in the record is off in Settings; the dialog would ask her to reword the frame or close the gap first.
“Signed as Dana Okafor, Clinical QA lead · recorded with name, date, time and meaning.”
The dashboard shows readiness rising week by week over the last 30 days to 84, against a target of 90 before the opening meeting; 94% of claims backed by the record; open gaps by inspection area; and back-room retrieval falling from 26:00 by hand in July to 3:12 in the September mock and 0:41 on Oct 5.
The same inspection, seen by the five people who carry it — what their preparation looked like, and what it looks like now.
Six specialist agents index the record, find topics, draft, check, rehearse and fetch documents; the QA lead approves every storyboard.
Reads the eTMF, CTMS, EDC, quality system, training and IRT records in place and labels each document by TMF zone, site, document date and filed date, flagging any filed after the inspection notice.
Finds the topics an inspector is likely to probe — important deviations, out-of-window visits, consent errors, excursions, system changes — ranks them and maps each to one of the nine inspection areas.
Drafts each storyboard in eight frames — the issue, how we found it, timeline, impact, root cause, what we did, proof it worked, where it stands — in short factual sentences, with dates and numbers exactly as the record gives them.
Re-reads every storyboard sentence and every rehearsed answer against its sources — supported, partly supported, or not in the record — and opens a gap with an owner for each one that falls short.
Plays the inspector: asks FDA- or MHRA-style questions one at a time, checks the answer sentence by sentence and gives at most four coaching points.
Finds the documents an inspector asks for, checks each is the effective version and complete, lists what is sent and keeps the request clock.
Approves, signs inspection-ready. The agents propose; a named person decides.
Ask about any storyboard, gap or likely inspector question — or say what to change. Every answer points to its source.
Where are we weakest for the FDA inspection?
Two areas are below 80 on the readiness radar. Protocol compliance is at 72 — in SB-01 the coordinator’s training is not in the record. Data handling is at 76 — in SB-05, 38 changes have no approval and the reopening memo is unsigned. Records & TMF is next at 80. Readiness is 84; closing the open high-severity gaps would lift it to about 88.
What will the inspector ask about visit windows?
Expect: how do you know the Site 114 corrective action worked; was any primary endpoint score lost; and how do you find out-of-window visits at other sites. The first is still not rehearsed — Dr. Maya Chen is the person who answers it.
Add a rule: flag every document filed after the inspection notice
Done. The new rule marks any cited document filed after Oct 5, 2026 and asks for a note-to-file. It already applies to 37 documents; 11 have notes. Saved to the audit trail.
Draft a storyboard for the IB v7 acknowledgement at Site 029
Started SB-13, IB v7 acknowledgement at Site 029, in Records & TMF. Six documents were found, including the site’s TMF completeness at 81%. The eight frames are being drafted; it is on the Storyboards list as Drafted, owner Omar Haddad.
12 screens from the working solution, on its sample data. Pick one to see it large.
The inspection, who signs and the rules every rebuild is checked against are settings, set once per inspection.
| Setting | Default | Choose from |
|---|---|---|
| Authority and type | FDA — sponsor inspection (BIMO) | FDA — sponsor inspection (BIMO) · MHRA — GCP systems inspection · EMA — product-related GCP inspection |
| Storyboard approval | Dana Okafor — Clinical QA | Dana Okafor — Clinical QA · Priya Raman — Clinical operations |
| Medical storyboards also need review by | Dr. Maya Chen | Dr. Maya Chen · Dana Okafor |
| Approve with an open gap | Off | On · Off |
| Every sentence needs a citation | On | Locked on — cannot be turned off |
| Partly supported when the source is a draft | On | On · Off |
| Flag numbers that differ from the clinical study report | On | On · Off |
| Warn when a cited document is superseded | On | On · Off |
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 study data · “target” = the design goal, measured in the live solution · “estimated” = our estimate · “typical” = published figures (MHRA GCP inspection metrics 2016–17: at least one critical finding at 7 of 15 commercial sponsors inspected). Inspection areas follow FDA Compliance Program 7348.810. People, companies, studies, sites and documents named on this page are fictional — characters and sample data in the working solution.
Software that gets a sponsor ready for an FDA or MHRA GCP inspection: the topics an inspector will probe, a factual storyboard for each, the gaps in the record, and rehearsed answers for the front room. GCP Inspection Readiness adds six agents that read the trial record in place, draft storyboards with a citation on every sentence, check each sentence against its source and play the inspector — the QA lead signs each storyboard.
Each storyboard is told in eight frames: the issue, how we found it, timeline, impact, root cause, what we did, proof it worked and where it stands. The Storyboard Writer drafts every sentence with a citation to a document and passage, and the Claim Checker marks each one supported, partly supported or not in the record.
The sentence is marked not in the record, with what the record says instead, and a gap opens with an owner and a due date. The owner finds the missing document or rewords the frame to what the record supports; the solution can propose a sentence the record fully supports.
Yes. The Inspector Simulator asks FDA- or MHRA-style questions from a question bank by inspection area. The person answers as in the front room, each sentence is checked against the record, and the cited model answer and up to four coaching points follow.
When the inspector asks for documents, the Back-room Retriever finds them, checks each is the effective version and complete, lists what is sent and keeps the request clock. A person releases every document. In the sample study’s Oct 5 mock the median was 0:41 per request.
No. The agents read the eTMF and the other systems in place and never move, edit or file a document. Documents filed after the inspection notice are flagged, and notes-to-file it drafts are filed only after the TMF manager signs them.
Yes. Agents propose topics, draft and check; a person decides which topics become storyboards, the QA approver signs each storyboard with the meaning of the signature, medical storyboards also need the medical monitor, and a person releases every document to the inspector.
The Agentic Solution Engine builds and deploys it from your requirements and documents — your inspection notice, your TMF plan and a sample of your study records — and it goes live once every quality gate has passed. We will walk you through it on one of your own studies first.
A checked first-draft protocol in days, in your approved language.
1-min demoeTMF Quality CheckAn inspection-ready trial master file every day, each site scored on what is due today.
1-min demoClinical Study Report AuthoringResults drafted within days of database lock, linked to tables.
1-min demoWe'll run GCP Inspection Readiness on one of your own studies and its trial master file.