R&D stage-gate governance that builds every gate dossier from cited evidence, on the same criteria every time
Gate dossiers built from live programme evidence in days, on the same criteria each time.
See one case, screen by screen ↓It is how a research organisation decides, at each milestone — target to hit, hit to lead, lead to candidate, candidate to IND — whether a programme gets Go, Go with conditions, Hold or Stop. R&D Stage-Gate Governance builds the evidence for each gate decision: six agents read the programme's documents, score it on the gate's criteria with every value cited, and draft the dossier, risks and conditions for the portfolio committee.
How to choose AI solutions for life sciences →
Updated
Before its first-in-human study, a research programme faces four gates: target to hit, hit to lead, lead to candidate, candidate to IND. At each one the programme team builds a dossier by hand from the target assessment, the pharmacology, in vivo, DMPK, toxicology and safety pharmacology reports, the route and solid-form report, Legal's IP memo, the competitive landscape brief and Finance's cost and timeline estimate — then turns it into slides for the portfolio review committee.
Each therapeutic area formats and weighs its criteria its own way, so a 72 in Neuroscience and a 72 in Oncology do not mean the same thing. Values go stale after the deck is frozen — a draft hERG result replaced by the final report, a cost estimate revised two days later. And the conditions the committee sets live in the minutes, where nobody sees that a 28-day monkey study report is four days late.
Estimated split for one lead-to-candidate dossier, by hand and with the solution.
Six agents read the science, CMC, IP, market and cost, score the gate's criteria with every value cited and draft the conditions; the committee votes and the chair signs.
Northwind Bio's Portfolio Review Committee meets on Wednesday, October 14, with four decisions and $73.5 M asked. One of them is NWB-4417, an oral NLRP3 inhibitor for hidradenitis suppurativa, at the lead-to-candidate gate. Here is what happens, screen by screen, in the working solution.
Omar chairs the committee; Dana Okafor, the portfolio governance lead, is its secretary. Four decisions are on the agenda: NWB-4417 at lead to candidate, NWB-2290 and NWB-8803 at candidate to IND, NWB-3105 at hit to lead. Thirteen of 20 votes are in and 18 of 20 pre-reads opened, with $73.5 M asked if all four items go ahead.
The agenda went to the five members on Oct 7 at 07:50, marked market-sensitive; dossiers for the next committees build ten days ahead.
The portfolio is one funnel: 12 programmes at four gates, each bubble sized by its cost to the next gate and coloured by its evidence score. He clicks NLRP3. The 86 is built dimension by dimension — weight, score and the document behind each. Safety carries 20% of the weight and scores 60, and the evidence is TX-4417-02.
“Show me why it is 86 — and what it has to prove before we spend $11.8 M.”
Under “Conditions to pass” the first chip reads “Exposure margin ≥ 10× before GLP”. It opens the 14-day rat dose-range-finding study, CRO draft report, page 11: at 300 mg/kg/day ALT rose to 3.1× the concurrent control in 4 of 10 animals; the NOAEL was 100 mg/kg/day, an exposure margin of 7.8× over the predicted human AUC. The changes were reversible after 14 days off dose.
Lead-to-candidate criteria v4.2: a margin of 5–10-fold is “partly met and needs a mitigation plan”.
The science reader reads the target assessment, pharmacology, in vivo, DMPK, toxicology and safety pharmacology reports. The CMC and IP reader reads the route and solid-form report and Legal's memo of Oct 2, and flags form B conversion. The market and cost reader reads the NLRP3 landscape brief and Finance's estimate. The criteria scorer applies template v4.2; the risk analyst and the dossier writer follow.
Committee: “4 of 5 votes in — waiting for the chair.”
hERG IC50 is now 42 µM in the final safety pharmacology report — the draft said 38. The cost to IND fell to $11.8 M from $12.4 M. Both are flagged beside the criterion, and the score stays at 86. Three criteria are partly met: the exposure margin, the solid form (form B converts at 60 °C / 75% RH) and competitive differentiation (six clinical rivals).
“Dossier rebuilt — 2 values changed since Oct 1 · score still 86”
Recommendation, programme summary, criteria, risks and mitigations, cost and value, decision requested — every number cited, and a click on a number opens its passage. It leads with Go with conditions, then the gap: a mild, reversible liver-enzyme rise at the top dose that a 28-day study can close before GLP work begins. Dr. Maya Chen checked and released it.
The dossier writer never expresses a vote or a decision — that belongs to the committee.
Each risk is rated for likelihood and impact and tied to its source. The liver-enzyme rise could narrow the clinical margin — Dr. Ines Moreau, a 28-day rat study with a liver panel. Form B converts in hot, humid storage — Priya Raman. A crowded class with six clinical-stage rivals — Lena Ortiz. A single supplier for the key starting material, a high placebo response in hidradenitis trials and a GLP start that could slip complete the register.
Lena Ortiz, Dr. Ines Moreau and Priya Raman voted Go with conditions; Sam Patel voted Go. The risk analyst has drafted three conditions with owners and dates: the 28-day rat study by Jan 15, 2027, the form B control strategy by Feb 26, 2027, and differentiation re-tested at the IND gate. Omar keeps all three and the drafted rationale for the record.
Dr. Ines Moreau: “7.8× is workable for a long-term skin indication only if the 28-day study confirms the liver change reverses.”
The signing screen shows the programme, the gate, the decision, three conditions and the votes — 3 Go with conditions, 1 Go, his completes 5 of 5. He chooses the meaning of his signature and enters his password. With all five votes in, the decision is taken by written procedure and noted at the Oct 14 meeting.
“The record is sealed when you sign.” … “A correction later becomes a new record that points to this one.”
PD-26-038 is sealed with its rationale, its three conditions — each open with an owner and a due date — the signer, the meaning of the signature and a record seal. The decision lands in the registry, and on the gate map NLRP3 moves to candidate to IND with its signed conditions beside it.
Signed by Dr. Omar Haddad · meaning: “I approve this portfolio decision and its conditions”.
The same decision, seen by the five people who carry it — what the weeks before a committee looked like, and what they look like now.
Six agents read the science, CMC, IP, market and cost, score the gate's criteria with every value cited and draft the conditions; the committee votes and the chair signs.
Reads target assessments, pharmacology, in vivo, DMPK, toxicology and safety pharmacology reports and extracts every value a criterion needs, with units, species, study and page. When two reports disagree it takes the newer final report and flags the older one.
Reads route, solid-form and stability reports and Legal's IP memos — route length, scale, polymorph risk, filing dates and freedom to operate.
Reads competitive landscape briefs and Finance's cost and timeline estimate: rivals by phase, the claimed differentiator, cost and months to the next gate, probability of reaching launch and risk-adjusted value.
Applies the gate's criteria template exactly as written — thresholds, partly-met bands and weights — marks each criterion met, partly met or not met, and computes the weighted score.
Turns partly-met criteria and source caveats into a risk register — likelihood, impact, owner, mitigation — and drafts conditions for the committee.
Drafts the six-section gate dossier — recommendation, programme summary, criteria, risks, cost and value, decision requested — for a committee that reads in ten minutes.
Votes, the chair signs. The agents propose; a named person decides.
Ask about any programme, dossier, vote or decision — every answer cites its source.
Summarise NWB-4417 for the committee
NWB-4417, an oral NLRP3 inhibitor for hidradenitis suppurativa, meets 11 of 14 lead-to-candidate criteria and scores 86. Potency is 18 nM in human whole blood with more than 300× selectivity, and the predicted dose is 120 mg once daily. The gap is safety: the exposure margin is 7.8× against a 10× criterion, from a reversible ALT rise at 300 mg/kg. The agents recommend Go with conditions — a 28-day rat study before GLP.
Which dossiers have a safety margin below 10×?
Three: NWB-8803, the Claudin 18.2 ADC, at 3× (candidate to IND); NWB-5521, the LRRK2 inhibitor, at 4.2× (lead to candidate); and NWB-4417, the NLRP3 inhibitor, at 7.8× (lead to candidate). NWB-8803's margin is at the highest non-severely toxic dose, NWB-5521 shows lung changes, and NWB-4417's liver change is mild and reversible.
What changed since the last committee?
Since the Sep 30 committee, two decisions were signed — NWB-6610 Go with conditions, NWB-0412 Go. In NWB-4417 two values changed: hERG IC50 is now 42 µM (was 38) and the IND estimate fell to $11.8 M (was $12.4 M); the score stays at 86. One condition is overdue — NWB-2204's 28-day monkey report, due Oct 3.
Why hold NWB-3105?
NWB-3105 has only one hit series and poor drug properties — solubility 8 µM with high clearance. Potency is good (DC50 22 nM). Sam Patel suggests a six-month, $1.1 M chemistry sprint as a condition instead of a plain hold.
13 screens from the working solution, on its sample data. Pick one to see it large.
The criteria each gate is scored on, who signs, and how the agents work are settings — versioned and on the audit trail.
| Setting | Default | Choose from |
|---|---|---|
| Criteria template in use | Lead → Candidate v4.2 | Target → Hit v3.1 · Hit → Lead v3.4 · Lead → Candidate v4.2 · Candidate → IND v2.8 |
| Exposure margin at NOAEL (Lead → Candidate) | ≥ 10× | A threshold per criterion; a change needs the template owner's approval |
| Who signs Lead → Candidate | Committee chair, full review — Dr. Omar Haddad | Dr. Omar Haddad · Lena Ortiz (deputy chair) |
| Who signs Candidate → IND | Committee chair, with Finance co-signing the budget | Dr. Omar Haddad + Sam Patel · Lena Ortiz (deputy chair) |
| Committee and programme team only | On | On · Off |
| Watermark every export | On — reader's name and time on every page | On · Off |
| Rebuild a dossier when new evidence lands | On — changes flagged for the programme leader | On · Off |
| Draft conditions from partly-met criteria | On — the chair edits and approves every condition | 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 portfolio · “target” = the design goal, measured in the live solution · “estimated” = our estimate · “typical” = common practice and published figures (2–4 weeks to build a gate dossier by hand is common practice; IND-enabling packages cost $2–7 M for a small molecule — Charles River). People, companies, programmes and studies named on this page are fictional — characters and sample data in the working solution.
It is how a research organisation decides, at each milestone — target to hit, hit to lead, lead to candidate, candidate to IND — whether a programme gets Go, Go with conditions, Hold or Stop. R&D Stage-Gate Governance builds the evidence for each gate decision: six agents read the programme's documents, score it on the gate's criteria with every value cited, and draft the dossier, risks and conditions for the portfolio committee.
The criteria scorer applies the gate's template exactly as written — for lead to candidate, 14 criteria across target, efficacy, drug properties, safety, manufacturing, IP and market, and cost and time. Each criterion is met, partly met or not met against its threshold, and the weighted score records the template version it was built on.
Yes. On the gate map the score breaks down by dimension, each with its weight, score and the document behind it, and every condition carries the code of the study it rests on. A click opens the cited passage with the values highlighted.
The dossier is rebuilt when new evidence lands, and values that changed since the last build are flagged for the programme leader. In the sample case the final hERG report replaced a draft value of 38 µM with 42 µM and the cost to IND fell from $12.4 M to $11.8 M; both were marked and the score stayed at 86.
No. The agents read, score and draft, and their recommendation is advisory. Committee members vote, signing opens once at least 2 of the 4 members have voted, and the chair signs with the meaning of the signature; at candidate to IND, Finance co-signs the budget.
Each condition the chair signs has an owner and a due date, is sealed into the decision record and is tracked in the registry. Open and overdue conditions show on the gate map and the dashboard, along with the share closed on time.
Dossiers are treated as market-sensitive: only the committee and the programme team can open them, every export is watermarked with the reader's name and time, exports cannot be emailed outside the company, and readers of a partnered programme's dossier are added to the insider list.
The Agentic Solution Engine builds and deploys it from your requirements and documents — your gate criteria, governance calendar and a sample of past gate dossiers and their evidence — and it goes live once every quality gate has passed. We will walk you through it on your own programmes first.
Sourced answers from papers, patents and reports in seconds, our own disagreements first.
1-min demoCompetitive Pipeline TrackingEvery rival trial, readout and approval in a sourced brief each Monday
1-min demoPreclinical Study Report AutomationA draft GLP report in hours, every number traced to raw data.
1-min demoWe'll run R&D Stage-Gate Governance on a sample of your own past gate dossiers.