IND nonclinical summaries for Module 2.6 and 2.4, drafted from the study reports with every number checked
The nonclinical summaries for an IND drafted in hours, with every number checked against its report.
See one case, screen by screen ↓The GLP reports land, the Module 2 sign-off date does not move, and someone has to turn thousands of report pages into written and tabulated summaries in the M4S headings of the house template. In the sample IND here that is 20 reports and 9,860 pages: the 28-day rat study alone runs to 2,914 pages.
Most of the effort is not writing. It is finding Table 14 on page 142, copying a value into the template, building the same numbers again into the tabulated summaries, then checking every one of them by hand — and redoing the exposure margins whenever a value moves. One transposed AUC changes the margin the starting dose rests on, and it is easy to miss in a 34-page toxicology summary.
Estimated split for the written and tabulated summaries of one first-in-human IND, by hand and with the solution, including author editing and review.
Six specialist agents read, plan, draft, tabulate and check every number; the writer and section owners resolve and sign.
NWB-4417 is an oral RIPK1 inhibitor heading for its first-in-human IND, 44 days out. Here is how its 2.6.6 Toxicology Written Summary gets from three disagreeing numbers to an approved section and a drafted 2.4 overview, screen by screen, in the working solution.
Lena opens the IND: 44 days to submission on Nov 20, Module 2 sign-off due Oct 30. Two of eight sections are approved, 2,318 numbers have been checked and 4 are left to resolve. A map wires each of the 20 study reports to the Module 2 sections it feeds.
“1,284 findings extracted, each with its page and table.”
TX-311, the 28-day GLP rat study with 14-day recovery and TK, is 2,914 pages. The reader extracted 286 findings from it — target organ, NOAEL, ALT, hypertrophy incidence, AUC at the NOAEL — each with its page, a confidence score and the report passage highlighted.
It never infers a NOAEL the study director did not state.
The draft opens with every sentence cited. One is red: ALT at 100 mg/kg/day is written as 1.8-fold in females. Beside it, Table 11 of TX-311 on page 52 is open, with a line from the sentence to the cell — 1.5-fold for females; 1.8 is the male value.
“Disagrees with the source. The draft repeats the male value.”
Lena takes 1.5 from the source and moves on. Next flag: the AUC0–24 at the dog NOAEL reads 22,400 ng·h/mL. Table 9 of TX-312 gives 24,200 (sexes combined). She chooses “Use the source value”; “Keep mine” is there, but sign-off stays blocked until the number matches.
The dog margin was built on the mistyped AUC: 22,400 ÷ 610 = 36.7, written as 37-fold. The number checker recomputes it from the source values — 24,200 ÷ 610 = 39.7, so 40-fold — and shows its working against PK-230: 79-fold to the rat NOAEL, HED 4.8 mg/kg, maximum recommended starting dose 29 mg.
“Number checker recomputed the dog margin: 37 → 40-fold.”
The number check for 2.6.6: 428 numbers in 186 statements — 367 exactly as reported, 61 derived and recomputed (margins, HED, fold changes), 0 disagree. Each row shows the value in the draft, the value in the source and where it sits in the report.
“Every number matches its source.”
Lena signs with her user ID and password. The meaning of the signature is “I drafted this section with the agents and checked every number against its source.” The AI use is recorded with it: drafted by the summary writer, 428 numbers checked by the number checker, 3 corrected from the sources.
The section goes to Dr. Omar Haddad, its approver. He sees Lena’s author signature and that every number matches, then approves with his own e-signature: “I approve the scientific content and interpretation of this section.” Every fix and both signatures are in the section history.
With 2.6.6 approved, the summary writer drafts the 2.4 Nonclinical Overview from the 2.6 sections and the PK-230 starting-dose rationale: 142 statements, every one cited. The number checker recomputes the margins and HED; the consistency checker compares the overview with each section and the Investigator’s Brochure v3.0.
Ready for Lena’s review — then Dr. Maya Chen approves.
Hours to a first draft by section against the typical hours by hand, statements linked to a source, numbers matched first time, review rounds per section — and what reviewers changed, by quality area. Reviewers spend their edits on emphasis and clarity; almost none on correctness.
The same IND, seen by the writer, the section approvers and the people who own the submission date.
Six specialist agents read, plan, draft, tabulate and check every number; the writer and section owners resolve and sign.
Reads each nonclinical study report — design, GLP statement, doses, findings tables, incidences, toxicokinetics — and records every finding with its study, species, sex, dose, day, value, unit, page and table.
Maps findings to the ICH M4S headings of the house template, lists the studies cited per section and flags any study that feeds no section.
Drafts 2.6.2, 2.6.4, 2.6.6 and the 2.4 overview in the house style, one cited sentence at a time.
Builds the 2.6.3, 2.6.5 and 2.6.7 tabulated summaries in the M4S formats from the same findings the text uses, each cell linked to its finding.
Compares every number in every section with its source and recomputes derived values — exposure margins, human-equivalent dose and the starting-dose maths.
Compares the sections with each other, with the Investigator’s Brochure and with the protocol, and reports conflicts with both citations.
Resolve, sign and approve. The agents propose; a named person decides.
Anyone on the IND team can ask in plain words, or tell it what to change. Answers cite the study reports they draw on.
What are the NOAELs, and what margin do they give the starting dose?
Two pivotal 28-day GLP studies set the NOAELs. Rat: 30 mg/kg/day, AUC0–24 48,200 ng·h/mL, a 79-fold margin at 10 mg. Dog: 15 mg/kg/day, AUC0–24 24,200 ng·h/mL, a 40-fold margin. Both margins use the projected human AUC of 610 ng·h/mL at the 10 mg starting dose.
Which numbers still disagree with the study reports?
Four. In 2.6.6: ALT fold change in females at 100 mg/kg/day, draft 1.8, source 1.5; AUC0–24 at the dog NOAEL, draft 22,400, source 24,200; the exposure margin to the dog NOAEL, draft 37, recomputed 40. In 2.6.4: human plasma protein binding, draft 97.8, source 97.2. I have opened the number check for 2.6.6 — one click takes each value from its source.
Add the in vivo micronucleus result to the 2.6.6 brief summary
Added to 2.6.6.1 Brief summary: “NWB-4417 was not genotoxic in a standard battery, including an in vivo micronucleus assay integrated into the 28-day rat study.” It cites the Ames test and TX-311 Appendix 12. The number checker confirmed the dose (100 mg/kg/day). The edit is in the section history.
How much time does this save?
The first draft of 2.6.2, 2.6.4 and 2.6.6 took the agents 2.6 hours for this IND (3.0 hours with the 2.6.7 tables). The published benchmark for drafting those summaries by hand is about 100 hours. Reviewers now spend their time on emphasis and clarity rather than finding Table 14 on page 142.
13 screens from the working solution, on its sample data. Pick one to see it large.
Template, citation style, number checking and who approves each section are settings, not a project.
| Setting | Default | Choose from |
|---|---|---|
| House template | Northwind nonclinical v6.1 | v6.1 · v6.0 · Partner template (ICH M4S plain) |
| Citation style | Report · page · table (inline) | Inline · Report number only · Footnotes |
| Units for exposure | ng·h/mL | ng·h/mL · µg·h/mL |
| Every number compared with its source | Always on | Sign-off blocked while any number disagrees |
| Recompute derived values | On | On · Off |
| Flag rounding (12.1% written as 12%) | Off | On · Off |
| Who approves each section | Named per section | Dr. Omar Haddad · Priya Raman · Sam Patel · Dr. Maya Chen |
| Second reviewer for 2.4 (the regulatory lead) | 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 IND data · “estimated” = our estimate · “typical” = published figures (IND-drafting study, arXiv 2509.09738, 2025: about 100 hours by hand for the 2.6.2, 2.6.4 and 2.6.6 written summaries) · Section numbering follows ICH M4S (CTD 2.4 and 2.6). People, companies and products named on this page are fictional — characters and sample data in the working solution.
It drafts the nonclinical sections of an IND’s Module 2 — the 2.6 written and tabulated summaries for pharmacology, pharmacokinetics and toxicology, and the 2.4 nonclinical overview — from your study reports, in your house template. Every sentence cites the report page and table it came from, and the writer and section approvers sign.
The number checker compares every number in every section with the value it cites, and recomputes derived values such as exposure margins, human-equivalent dose and the starting dose. A number that disagrees is shown next to the source value, and sign-off is blocked until it matches or the author records why it differs.
No. The study report reader never infers a NOAEL the study director did not state, and the summary writer makes no adverse or non-adverse calls the report does not make. Emphasis and judgement are left to the author and the section approver.
Yes. The agents write into your template with its ICH M4S headings, which they may not rename. Citation style and exposure units are settings, and each section exports to Word in the template with a citation map.
Add the new version and it is re-read and compared with the previous one. The statements that cite a changed value are found across all sections and sent back to number check; the writer decides on the changes.
Yes. The agents draft, build tables and check, but never sign. The writer signs each section as author, then its named approver — the study toxicologist, the DMPK lead, the pharmacology lead or the head of nonclinical safety — approves with an e-signature that records its meaning.
Yes, on every section: which agent drafted or checked what, how many numbers were checked and how many were corrected from the sources, kept with the section for your GenAI policy file.
The Agentic Solution Engine builds and deploys it from your requirements — your house template, prior IND summaries and a set of study reports — and it goes live once every quality gate has passed. We will walk you through it on your own reports first.
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 a Module 2.6 section from a set of your own nonclinical study reports.