We score your NIH application against the real review framework — without writing a word of your science.
Most grant software does one of two things: it finds you money, or it drafts your proposal. This does neither. It reads what you already wrote, scores it against a published rubric built on NIH's Simplified Review Framework, and tells you — specifically, with citations — what a fair panel would object to and what to fix first.
Against a fixed, published rubric
Ranked, cited, attachment-level findings
Your ideas stay provably your own
Advisory readiness signal — not a prediction of NIH's funding decision.
— WHERE WE FIT
An honest map of the grant-software space — and the one bucket we chose.
Three things get sold to grant writers. They are not substitutes for each other, and pretending otherwise is how teams end up with a polished document nobody scored. We name the categories, not the companies.
Discovery tools
Opportunity search, deadline feeds, funder databases. They answer 'what could I apply to?' Useful, and upstream of everything we do — we index a funding directory ourselves for exactly this reason.
AI drafting tools
Generative assistants that produce Specific Aims, Research Strategy prose, or full narrative sections from prompts and prior text. They answer 'what should the page say?' by writing it for you.
Scoring & diagnosis
Read the applicant's own text, score it against the published review framework, run deterministic compliance gates, and rank what to fix. Answers 'is this competitive yet, and what is holding it back?'
— THE POLICY WEDGE
“We don't touch your science” stopped being a limitation. NIH made it an advantage.
The rules changed underneath the drafting-tool category. Here is what NIH has actually published — read it yourself, every source is linked.
Applications substantially developed by AI are not treated as the applicant's original ideas.
NIH's notice on the use of generative AI in application development states that NIH does not consider applications — or sections of them — substantially developed by AI to be the original ideas of the applicant, and that such applications may be referred for further review as possible research misconduct. The policy direction set in this notice now carries forward into the NIH Grants Policy Statement effective March 2026.
Peer reviewers are prohibited from putting your application into a generative AI tool.
NIH prohibits scientific peer reviewers from using generative AI technologies to analyze or critique applications, because doing so breaches the confidentiality of the review process. Your application is confidential pre-publication IP on NIH's side of the table too — which is exactly the standard we hold ourselves to on ours.
Reviewers say they can tell — and say it costs applicants at triage.
Reviewer-side research and commentary has reported that a majority of applications reviewers identified as AI-written did not progress past initial triage. We are repeating a reported finding, not producing a measurement of our own: we have not run that study, we do not have the underlying cohort, and we will not attach a precise percentage to it. Treat it as directional evidence that reviewer-facing AI prose carries real downside risk.
A tool that drafts your proposal walks you toward an originality problem you now have to defend. A tool that scores your own writing without generating any of your science leaves your ideas provably yours. That is not a feature gap on our side — it is the structural reason to use us.
WE SCORE · WE DIAGNOSE · WE CHECK FORMAT · WE NEVER WRITE YOUR SECTIONS
— WHAT MAKES THE SCORE DEFENSIBLE
“An AI reviewed it” is not a finding. A traceable score is.
Improvised reviewer personas can produce sharp prose and a different answer every time you ask. That is a smart demo and a weak instrument. Every number we hand you traces back to a published rubric line plus a cited source fact — which means you can argue with it, and so can your co-investigators.
A published standard, not a mood
The rubric is written down, versioned, and mapped to NIH's Simplified Review Framework — Factor 1 (Importance), Factor 2 (Rigor and Feasibility), Factor 3 (Expertise and Resources, rated sufficient or not). Same application, same rubric, same reading. You can read the whole thing before you pay.
Read the methodology →Winner Intelligence runs on NIH RePORTER records
A curated, growing set of real NIH awards — organization, PI, institute, activity code, fiscal year, dollars, abstract — each row linking back to its RePORTER source. When we tell you who has been funded on this mechanism, you can click through and check it.
Open the explorer →Some answers should never involve a model
Page limits by attachment and activity code, SBIR/STTR budget guideline levels, and the STTR work-split requirement (at least 40% small business, at least 30% research institution) are rules-table checks. They fire the same way every time, with the source rule cited, because arithmetic and published tables are not judgment calls.
See the page-limit matrix →The Submission Readiness Check uses no AI at all
A PDF parser and a rules table measure pages, margins, and font size against NIH format requirements, then the file is discarded when the check returns. No model call is involved in that product, so there is nothing about it to take on faith.
Run a format check →Where judgment genuinely is required — reading whether an aim is mechanistically sound — a language model does the reasoning against our rubric, and we say so plainly rather than dressing it up as a measurement. Scores are an advisory readiness signal, never a prediction of NIH's funding decision.
— FOUR INDEPENDENT SIGNALS, ONE DECISION
Four reads that do not depend on each other — so agreement means something.
Each one answers a different question with a different method. When they agree, you write. When they disagree, you know exactly where the argument is.
Submission Match
A mechanism-specific rubric — one track for academic R-series, one for SBIR/STTR — scored against your own text, with deterministic gates for registrations, budget guideline levels, and the data management plan. Returns a readiness range with the reasoning behind it, not a single false-precision number.
The cheapest correction in grant writing is finding out you are on the wrong mechanism before you write 12 pages for it.
Mock Study Section
Three reviewer personas score Factor 1 and Factor 2, rate Factor 3 as sufficient or not, and return a reviewer-style Overall Impact readiness range as a holistic judgment — plus strengths, weaknesses, and a fix list ranked CRITICAL, HIGH_IMPACT, POLISH. Unlimited rounds, including A1 responses.
You get Reviewer 2's objection while there is still time to fix it, instead of months later in a summary statement.
Winner Intelligence
A curated, growing set of real NIH awards from RePORTER, filterable by institute, activity code, fiscal year, disease area, and state — every row citable back to its source record, with CSV export.
If nobody resembling you has been funded by this institute on this mechanism, that is the most useful no you will get all quarter.
Funding Opportunity Directory
Verified non-dilutive and venture-philanthropy programs beyond NIH — disease foundations, state agencies, translational funds — each with real terms (equity, royalty, milestones) and a source page you can check.
A NO-GO on one mechanism is not a NO-GO on funding. It is a redirect.
— TRUST AS PRODUCT
For a scientist audience, the disclosure is the feature.
You evaluate instruments for a living. So we built this the way a methods section gets built: published standard, cited inputs, disclosed processing, stated limitations. Anything we cannot substantiate, we do not say — including about ourselves.
- PUBLISHED RUBRIC
The scoring standard is readable before you buy, and versioned when it changes.
- CITED FACTS
Page limits, budget levels, deadlines and award rows carry a source and a verification date.
- DISCLOSED PROCESSING
Sub-processors are named, retention is stated, deletion is a button with a receipt.
- STATED LIMITS
We say what this cannot do: it is not the real study section and it does not validate outcomes.
- NO FUNDING CLAIMS
No odds, no payline predictions, no implied guarantee. Readiness signal only.
- NO GHOSTWRITING
We never generate your Aims, Strategy, or narrative. Your science stays yours.
Target Intelligence Report on any target you choose.
A custom ~20-page cited scientific dossier on the target, mechanism, or disease area of your choice. Hand-prepared and delivered by email within 24 hours of purchase. One per Application Pass, LAUNCH50 sale only.
Everything we know, free.
NIH Submission Rules Checklist
Every rule that can get your application withdrawn — sourced and dated.
Open →EXPLAINERNIH Review Criteria (2025)
Factor 1, Factor 2, Factor 3, and how Overall Impact really gets set.
Open →REFERENCESBIR/STTR Budget Caps
Phase I, Phase II, CRP — guideline levels with fee, indirect, and IC-waiver context.
Open →MATRIXNIH Page-Limit Matrix
Every attachment, every activity code — with SBIR/STTR rows highlighted.
Open →GLOSSARYNIH & SBIR Glossary
NOFO, JIT, RPPR, MPI, Payline, ESI, UEI — plain-English definitions.
Open →GUIDENIH Resubmission (A1) Guide
One shot, 37-month clock, 1-page Introduction — the rules that actually apply.
Open →CHECKLISTJust-In-Time (JIT) Checklist
Other Support, IRB/IACUC, training certs — everything NIH asks for pre-award.
Open →.ICS FEEDDeadline Calendar
SBIR/STTR standard dates, adjusted for weekends and federal holidays.
Open →LOOKUPActivity Code Lookup
R43, R44, R41, R42 — page limits, aims, commercialization plan, deadlines.
Open →PLANNERRegistration Runway Planner
SAM.gov, eRA Commons, Grants.gov — every date backward-planned from your deadline.
Open →IC GUIDENIH Institutes & Centers
The major SBIR/STTR ICs — focus areas, clinical-trial policy, and small business links.
Open →OMNIBUS 2026What changed in the 2026 omnibus
One Clinical Trial Optional pathway, higher IC-specific ceilings, no late applications, and the new foreign-subaward bar.
Open →Find out if your next application is worth writing.
Registrations alone can take six weeks or more, and there is no deadline extension to fix errors. Score it now, while the calendar is still on your side.
Five minutes · No card · No success fee