Cardinal — Clinical Protocol Intelligence

Every avoidable amendment was once a design decision nobody checked.

Cardinal checks them — against 587,000 real trials — before you write a single section.

Median cost of one substantial amendment: $141K in Phase 2, $535K in Phase 3. Source: Tufts CSDD.
Cardinal
DESIGN CHECK · NSCLC · PHASE 2
Flagged before authoring
Your eligibility criteria sit in the top quartile for restrictiveness in this indication.
Eligibility and demographic changes drive 53% of substantial amendments.
Enrollment feasibility N = 280
P10P50P90
67th percentile · 95% CI 55–79 · n = 3,131 matched trials
587KTrials compared
3Risks flagged
0Sections written
Checked before drafting — not after review
Your data never trains any model
Encrypted in transit and at rest
ICH M11 · 21 CFR Part 11 aware
Full audit trail on every output
FDA · EMA · Health Canada
76%
Of trials now require at least one protocol amendment — up from 57% in 2015
45%
Of those amendments are judged avoidable — traceable to protocol design quality
53%
Start with eligibility criteria or study population — the decisions Cardinal checks first
$535K
Median direct cost of a single substantial Phase 3 amendment
Source: Tufts Center for the Study of Drug Development.

Where does your enrollment
target actually stand?

Three parameters — indication, phase, N. Nothing confidential, nothing uploaded. Benchmarked live against the trial registry. You will have an answer before you finish your coffee.

Cardinal · Design Check Awaiting parameters
Runs entirely in your browser. Nothing is transmitted or stored.
Enrollment feasibility
P10P50 (median)P90
See the full design check.
Enrollment is one of nine dimensions. Enter your name and work email to unlock eligibility burden, endpoint precedent, and amendment-risk profile — shown here instantly.
We will also reserve a free protocol assessment for you. No newsletter, no sequence.

Full design check —

·

Eligibility burden
Endpoint precedent Established
Amendment risk profile

Want Cardinal to run the remaining five dimensions against your actual protocol?

Claim your free assessment →

The expensive mistakes
are made before anyone writes.

An amendment is never really a documentation failure. It is a design decision — an eligibility rule, an endpoint, a sample size — that no one had the evidence to challenge at the time. By the time it surfaces in enrollment or review, it costs six figures and three months to undo.

i.

Decisions made on instinct

Sample size, eligibility, and endpoint choices are set by convention and committee memory. The precedent that would test them sits in a registry nobody queries mid-design.

ii.

Contradictions hide in plain sight

A sample size stated one way in the objectives and another in the statistics. An endpoint defined twice, differently. Manual review misses them; reviewers do not.

iii.

The real reviewer is the first test

Hard questions from biostatistics, ethics, and regulatory arrive for the first time in an actual review — the most expensive possible place to discover a flaw.

Evidence at the moment
of the decision.

Upload your Investigator's Brochure. Cardinal returns a complete, ICH M11-aligned draft — then keeps working: benchmarking your design against precedent, checking the document against itself, rehearsing your review, and generating everything downstream.

  • Every clinical claim traced to its source in your IB — never fabricated.
  • Eligibility, endpoint, and enrollment choices benchmarked against similar trials from ClinicalTrials.gov, with a confidence range attached to every score.
  • Contradictions across sections, and amendment-prone patterns, flagged automatically as you write — not discovered at review.
  • Six expert review personas rehearse the real review before you submit — clinical, biostatistics, operations, data standards, ethics, and regulatory.
  • SAP, ICF, synopsis, and EDC specification generated directly from your protocol — one source, no re-typing. Amendment documents generated in one click from flagged risks, ICH E6(R2) and 21 CFR 312.30 compliant.
  • Editable Microsoft Word output that drops into your existing review workflow.
Cardinal · Generation Pipeline
Investigator's Brochure ingested · 47 fields extracted
0:32
Indication precedent loaded · 24 reference trials
1:18
Section structure planned · 18 sections
1:45
Schedule of Assessments generated
2:51
Statistical considerations drafted
3:34
6
Cross-section consistency check running...
4:12

Design it, write it, prove it —
one platform, start to finish.

Cardinal is built around the workflow your medical writers and clinical operations team already know — not a new paradigm to learn. It just doesn't stop at the draft.

i

Design with evidence

Compare design options in Design Studio before a section is written — sample size, eligibility, and endpoint choice benchmarked against thousands of matched trials.

ii

Draft and validate

Upload your IB and Cardinal drafts the full protocol, while Quality Scoring and the Consistency Engine check every section — and each other — as you go.

iii

Rehearse and generate

Reviewer Simulation raises the questions a real review would, Amendment Risk flags what's likely to cost you later, and the SAP, ICF, synopsis, and EDC spec generate straight from your protocol.

Precedent Intelligence PHASE 2 · RA · MTX-IR
Endpoint suggestion
72% of recent Phase 2 RA trials in MTX-inadequate responders use ACR50 at Week 12 as primary. ACR20 use has declined to 14% since 2022.
High confidence
88%
Design pattern
Median sample size for similar Phase 2b JAK1-selective inhibitor trials: n=348 across 3 arms, with 1:1:1 randomization.
High confidence
92%
Safety framework
JAK inhibitor class typically defines 7 AESIs: serious infections, herpes zoster, MACE, VTE, malignancy, GI perforation, hepatic injury.
Class consensus
95%

Precedent intelligence,
at the moment of authoring.

Cardinal does not generate protocols from a blank slate. Every endpoint suggestion, design choice, and safety framework is grounded in the precedent of similar trials — this is what powers Design Studio, surfaced from a structured index of past clinical research.

When ACR20 was the standard primary endpoint for Phase 2 RA trials, Cardinal would suggest it. Today, ACR50 has overtaken it — and Cardinal reflects that shift, with the data, the evidence grade, and the confidence range to back it up.

This is the difference between a writing tool that writes for you and a clinical development intelligence platform that thinks alongside you — and never makes the final call for you.

The complete protocol
intelligence suite.

Cardinal is not a writing assistant. It designs, drafts, validates, and pressure-tests the protocol — then generates everything downstream of it. Every capability is live; every output is traceable.

Design Studio

Compare two or more design options before authoring begins — sample size, eligibility, and endpoint choice, each benchmarked against a matched cohort of real trials with a confidence range attached.

IB-grounded generation

Mechanism, dosing, PK/PD, and safety pulled verbatim from your uploaded Investigator's Brochure. Never fabricated, never approximated.

Quality Scoring

Every section scored 0–100 on completeness, clarity, and regulatory alignment — recalculated automatically on every save.

Consistency Engine

Catches cross-section contradictions before they reach review — sample size, treatment duration, and endpoint definitions checked against themselves everywhere they appear.

Amendment Risk Engine

Scores overall amendment risk and surfaces specific flags across five dimensions — eligibility, endpoints, statistical design, schedule of assessments, and operational burden. Every flag includes the protocol evidence and a concrete recommendation.

Reviewer Simulation

Six expert review personas — clinical, biostatistics, operations, data standards, ethics, and regulatory — read the protocol and raise the questions a real reviewer would ask first.

Amendment Generation

From flagged risks to a complete, ICH E6(R2) / 21 CFR 312.30-compliant Protocol Amendment document in one click. Disputed flags are pre-populated and locked; accepted flags are excluded. Editable Word output ready for expert review.

Regulatory Intelligence Panel

Current FDA and EMA guidance, ICH guidelines, and precedent cases, surfaced automatically against whichever section you're writing — updated monthly.

Historical Benchmarking

The full distribution — not just a single score — for enrollment size, duration, and screen-failure rate, filterable by phase, sponsor type, region, and year.

Protocol Versioning

Upload a revised protocol for the same drug and indication and Cardinal automatically links it as v2, v3, and so on. A visual score trend shows how amendment risk changes across each revision — so you can see immediately whether changes improved the picture or introduced new exposure.

Protocol Intelligence Chat

Ask any question about an uploaded protocol — eligibility rationale, Schedule of Assessments design, statistical justification. Every answer is grounded in the actual protocol text and its risk flags, not general knowledge. Validated across 40 clinical question types by persona; scored 4.08/5.0 for protocol-specificity.

Reference Protocol Library

Upload competitor or historical protocols as benchmarks alongside your own. Cardinal keeps them in a separate library, scoped entirely to your organization — no other team can see or query your library. Compare risk profiles across programs without ever exposing sponsor data.

Every new trial should begin
knowing what every trial before it learned.

Half a million trials have already been run. Their designs, their endpoints, their failures — all of it public, and almost none of it available at the moment someone sits down to design the next one.

Merquell exists to close that gap. Not by replacing medical writers and clinical operations teams, but by giving them a platform that checks the decisions, not just the prose.

Company

Merquell Technologies Inc., federally incorporated in Canada. Headquartered in Toronto, Ontario.

Stage

Working with a select group of design partners to validate Cardinal and shape the roadmap ahead of general availability.

Who we build for

Sponsors and CROs designing and authoring clinical trial protocols — from first-in-human through Phase 3, at emerging biotechs and established organizations alike.

Regulatory focus

ICH M11 compliant generation. US FDA, EMA, and Health Canada awareness built into the core architecture.

A platform purpose-built for
regulated industries.

Cardinal never ranks or recommends an option, and never decides for you. It shows the evidence, names its source, and leaves the call — and the sign-off — with your team. Every score traces back to a dataset, a cohort size, and a confidence range.

Your data stays yours

Your Investigator's Brochures and protocol inputs are never used to train any language model, and are accessible only to your organization. Confidentiality is the default, not a setting.

Data security

Your IB and protocol inputs are encrypted in transit and at rest, processed under strict data agreements with the enterprise infrastructure that powers generation. AES-256 at rest.

Auditable output

Every generated section traces back to its source — IB extract, precedent reference, or user input. Designed for regulatory traceability from day one.

Multi-region ready

Built with awareness of FDA, EMA, PMDA, and Health Canada regulatory frameworks. Generation logic adapts to regional submission requirements.

Row-level org isolation — not just access control

Every database query is scoped to your organization at the row level. A CRO managing 40 sponsors cannot expose one sponsor's protocol intelligence to another — not through a misconfigured permission, not through a shared session, not through an indirect query. The architecture enforces it; policy alone does not. If you are one organization, you see only your data. Full stop.

Run Cardinal against
your protocol — at no cost.

We run Cardinal against one of your live protocols and walk you through everything it flags — design risks, contradictions, and the questions a reviewer would ask. Under mutual NDA, before anything is shared.

Mutual NDA signed before you share anything
Your design benchmarked against matched trials in your indication
Every amendment risk in your draft, flagged and explained
A sample SAP, ICF, and synopsis generated from your protocol
Thirty minutes with the founder. No sales team, no sequence.

Mutual NDA first. Thirty minutes with the founder — no sales team, no email sequence.

Assessment reserved.

You will hear from Pavan.kumar@merquell.com within two business days, with a mutual NDA and a link to book your thirty minutes.

Nothing else — no newsletter, no drip sequence.