CTD Module 3 quality sections, drafted from your development data with every number checked against its source
Module 3 quality sections drafted from your development data in about half the time, with every number checked against its source.
See one case, screen by screen ↓A Module 3 section is mostly numbers that already exist somewhere else: release results on six certificates of analysis, limits in the current specification, time points in a stability report, wording in the last approved filing. The CMC writer's job is to bring them together in the house template, in the right version, before the content lock.
Most of the hours go on copying, not writing. Results are retyped from certificates and LIMS exports, prior text is pasted and adapted line by line, and every table is checked by hand against its source. One pasted range from an earlier product, one raw instrument value instead of the certified result, or one limit from a superseded specification is enough to send a section back — and when a specification changes, someone has to find every section, in every filing, that still quotes the old value, including the Quality overall summary.
Our estimate of how the hours for one Module 3 section split, by hand and with the solution. The totals match the 13.5 h and 5.8 h targets.
Agents read the sources, pull the data, find approved wording, draft each section and check every number; the author edits and reviewers sign.
The Veltrimab BLA locks content on Oct 30. Section 3.2.P.5.4 Batch analyses is drafted and technically reviewed — but the number check has found two problems. Here is what happens next, screen by screen, in the working solution.
Maya opens the solution and sees Module 3 for three open filings: the Veltrimab BLA at 34 of 45 sections approved with content lock in 23 days, the Veltrimab MAA at 14 of 45, and a Corventa new-site supplement at 1 of 12. 3,912 numbers in the BLA are linked to a source, and 7 fixes stand between it and the lock.
Agents at work: “Change tracker found 4 places that quote the old pH limit in 3 sections.”
Priya Raman wrote the section; Sam Patel signed the technical review on Oct 6, subject to the number check. The Number checker read all 96 numbers in 38 seconds. Two do not match their sources, and Approve and sign stays greyed out until they do.
“2 numbers to fix before approval. Click a highlighted number to see its source.”
The text says SE-HPLC monomer ranged from 98.6% to 99.4%. The lowest result in the table is 98.3%, for PPQ batch VLT-DP-26P02. The Prior filing tab shows where the 98.6% came from: the approved Corventa 3.2.P.5.4 text, reused at 91% with every change marked, and pasted in on Oct 6.
Finding: “Range does not match the table.”
In Table 3.2.P.5.4-2, pH for VLT-DP-26P03 shows 6.15 from the LIMS export. The certificate of analysis, open beside the draft, reports 6.2 — and house style 4.2 reports results to the specification’s decimals. A third item is only a note: certificate VLT-DP-25A02 moved to revision 2 on Sep 22, results unchanged, and the citations already point to it.
Each finding carries its fix: “Use 98.3%” and “Use 6.2”. Maya applies both. The check turns clean — 96 of 96 numbers match their sources — and the remaining checks still hold: limits match 3.2.P.5.1 at specification version 4, units match the certificates, batch numbers and dates match the LIMS export.
“All numbers match their sources — ready for approval.”
The signing screen shows the section, the filing, version 0.4, 96 of 96 numbers matched and Sam’s technical review. Maya picks the meaning of her signature — “Approved for submission” — enters her password and confirms she reviewed the section, its number check and its sources. The signature is bound to version 0.4, and the activity trail records both fixes and the signature.
“Signed — 3.2.P.5.4 approved · added to the audit trail”
The 24-month interim stability report STB-DP-031 was released on Oct 5, so 3.2.P.8.1 can be drafted now. The Source reader finds it, the Data extractor reads 1,326 results for 3 batches and 2 conditions, the Section drafter starts from the approved Corventa 3.2.P.8.1 text and writes four paragraphs, a monomer trend figure and a stability table. The Number checker checks 46 numbers: all match. The section goes to Lena Ortiz for technical review.
Proposed shelf life: 24 months at 2 – 8 °C, protected from light — the drafter never claims a shelf life the data do not support.
Specification SPC-VLT-DP-004 version 4, effective Sep 29, narrowed pH from 5.8 – 6.4 to 5.9 – 6.3 under change control CC-2026-118. “Where used” lists six places across both Veltrimab filings: three are current, three still quote 5.8 – 6.4 — 3.2.P.5.6 and the Quality overall summary in the BLA, 3.2.P.5.1 in the MAA.
“The Change tracker can draft the update in each, for the author to confirm.”
3.2.P.5.6 Justification of specification(s) opens with the reason: the section still quotes the version 3 limit in 2 places, in the text and in Table 3.2.P.5.6-1. Both are listed as 5.8 – 6.4 → 5.9 – 6.3, with the change history on the specification page beside the draft. One click on “Apply 2 updates” puts them in; the section then goes back through the number check before anyone signs.
Sections drafted by agents and by hand each week, writing time per section type, what the Number checker caught — copied or typed wrong, decimals or rounding, superseded source, range not matching the table, units, summary out of sync — and readiness by filing against each content lock.
The same filing, seen by the people who carry it to content lock — what their weeks looked like, and what they look like now.
Agents read the sources, pull the data, find approved wording, draft each section and check every number; the author edits and reviewers sign.
Reads development reports, certificates, specifications and prior filings; maps each to the CTD sections it supports and records its version, effective date and approval status.
Extracts batch results, specification limits and stability results, each value with its document, version, page and confidence.
Finds the closest approved prior filing for each section, so the draft starts from wording already accepted, and shows every change against it.
Writes each Module 3 section in the house template, starting from the closest approved prior text and filling every table from the extracted data.
Checks every number in a draft against its source: value, unit, decimals, source version, the current specification, and the same value elsewhere — including the Quality overall summary.
Watches for new certificate revisions, stability pulls and specification versions, finds every section in every open filing that uses a changed value, and drafts the update.
Edits, then reviewers sign. The agents propose; a named person decides.
The CMC team can ask in plain words — or tell it what to draft, check or update. Every answer points to its source.
Why did the check flag the batch analysis section?
Two numbers in 3.2.P.5.4 do not match their sources. The text says monomer ranged from 98.6%, but the lowest result is 98.3% for VLT-DP-26P02 — 98.6% was pasted from the approved Corventa text. And pH for VLT-DP-26P03 shows the raw 6.15 from the LIMS export, while the certificate reports 6.2. Each finding has a one-click fix.
Update every section that still quotes the old pH limit
Specification version 4 narrowed pH to 5.9 – 6.3. The Change tracker drafted the update in 3 sections across 2 filings: 3.2.P.5.6 and the overall summary (Veltrimab BLA) and 3.2.P.5.1 (Veltrimab MAA). Each is now Drafted and needs a number check before signing.
What shelf life do the data support?
The 24-month data support 24 months at 2 – 8 °C, protected from light. Monomer fell on average 0.4% over 24 months (slope −0.017% per month) and stayed at or above 98.6% against a 95.0% limit. The overall summary still says 18 months — it is flagged to update.
Is the overall summary in sync?
Not yet — three values in Module 2.3 differ from Module 3: the pH criterion, the lowest monomer result (98.6% → 98.3%) and the shelf life (18 → 24 months).
13 screens from the working solution, on its sample data. Pick one to see it large.
The house template, the checks the Number checker applies, the technical reviewer for each area and the regulatory approver are all chosen in Settings.
| Setting | Default | Choose from |
|---|---|---|
| House template | Northwind Module 3 house template v5 | Northwind Module 3 house template v5 · Northwind Module 3 house template v4 · EU variation template v2 |
| Reuse approved wording | On — start from the closest approved prior filing | On · Off |
| Report results to the specification’s decimals | On — flag raw values that differ from the certified result | On · Off |
| Ranges in text must match the table | On | On · Off |
| Cite only current source versions | On | On · Off |
| Keep the overall summary in sync | On — compare Module 2.3 with Module 3 on every change | On · Off |
| Block approval while numbers are open | On | On · Off (a reason is then required to sign) |
| Regulatory approval | Dr. Maya Chen | Dr. Maya Chen · Dana Okafor |
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 filings · “target” = the design goal, measured in the live solution · “estimated” = our estimate · Context: ICH M4Q(R1), ICH Q6B and Q1E; industry pilots of AI drafting report up to 50% faster writing. People, companies and products named on this page are fictional — characters and sample data in the working solution.
Software for CMC regulatory writers that drafts CTD Module 3 quality sections — drug substance, drug product, appendices and the Quality overall summary — from your development reports, certificates of analysis, specifications and stability data. Agents draft each section in your house template and check every number against its source; authors edit, technical reviewers and the regulatory approver sign.
The Number checker finds the source of every number in the draft and compares value, unit and decimals. It checks that ranges in the text match the table, that limits match the current specification, that no superseded source is cited and that the Quality overall summary says the same. Each finding comes with a proposed fix.
Yes, if you want it to. The draft starts from the closest approved prior filing, replaces every fact and number with the new product’s, cites it, and shows each change against the prior text. You can turn reuse off in Settings.
The Change tracker finds every section in every open filing that quotes the changed value — Module 3 and the Quality overall summary — and drafts the update. A person applies it, and the section goes back through the number check before it can be signed.
Process and formulation development reports, characterisation and validation reports, certificates of analysis, LIMS batch analysis exports, specifications and change controls, stability reports, comparability protocols, prior approved submissions and your house style guide.
Yes. Agents draft and propose; people decide. Only a person applies a fix or an update, each section needs a technical review and then a regulatory approval with an electronic signature, and agent-written paragraphs are marked in the draft.
Yes. The working solution carries a BLA, an MAA and a new-site prior approval supplement side by side, each with its own Module 3 map, content lock and readiness. When a shared source changes, the Change tracker finds every open filing that quotes the old value.
The Agentic Solution Engine builds and deploys it from your requirements and documents — your house template, style guide, sample development reports and a prior approved filing — and it goes live once every quality gate has passed.
Ask the literature in plain words and get a sourced answer in seconds.
1-min filmPipeline WatchEvery rival trial, readout and approval lands in a sourced brief for each programme, every Monday.
1-min filmPreclinical ReportsRaw study data becomes a draft report in hours, with every number traced to its source.
1-min filmWe’ll draft and check a Module 3 section from a sample of your own development data.