Unsubmitted Specific Aims are pre-publication IP. We don't use yours.
Unsubmitted Specific Aims are pre-publication intellectual property. We don't use them to train models — ours or anyone else's. We calibrate against two things instead: whether the companies we scored actually got funded, and a benchmark set of published, awarded grants — which a reviewer advisory panel of former study-section members, now being assembled, will score independently.
The Submission Readiness Check involves no model at all. It's a PDF parser and a rules table. Your application is never sent to an AI.
Outcomes and a reviewer advisory panel. Not a scraped corpus.
- Outcome tracking. Every score we produce is labeled, over time, with whether the application it evaluated was funded. That is the signal — not the text of your application.
- Reviewer advisory panel. We are seating a panel of former NIH study-section members to score a benchmark set of published, awarded grants; once seated, we will measure our engine against their consensus and name members here.
- Public awarded data. NIH RePORTER abstracts are public. We ingest them for Winner Intelligence and for grounding the competitive position dimension of Agency Fit.
Six commitments. Each one costs us something.
Every claim on this site is sourced. Every rule is dated. If any of the below stops being true, this page is wrong and we're in trouble — that's the point.
LAST FULL REVIEW: 09 JUL 2026
Costs of professional and consultant services performed under a contingent fee basis, where the fee is contingent on the award of a Federal contract, are unallowable. Read as: we cannot legally take a percentage of your award. And we wouldn't want to.
Every rule cites its source, with a verified-on date.
The Submission Readiness Check involves no AI. It's a PDF parser and a rules table. Every rule links back to the specific NIH page — Format Attachments, GPS 2.3.9, an active NOFO — and shows the date we last verified the source. If the date is stale, so is the rule; we say so.
We re-verify quarterly. If a rule changes between passes and we ship an outdated check, we say so publicly in the change log.
No success fee, ever. Fixed price before we know your outcome.
Contingency fees from federal award funds are unallowable under FAR 31.205-33(f). Beyond that, a success fee gives us a reason to tell you to apply when you shouldn't. Our incentive is to tell you honestly when the answer is no.
When we don't believe an application will fund, we say so — even to a paying customer. That's the cost of being useful.
We won't publish a success rate we can't verify.
You can't measure a grant success rate from application data alone — NIH publishes award outcomes 6–18 months after submission, and success is confounded by resubmission behavior, institute payline shifts, and NOFO churn. Any number a competitor publishes without a linked outcome cohort is invented.
When we can verify a cohort, we'll publish the numbers with the methodology, sample size, and confidence interval. Not before.
We refuse work we can't do well.
We do NIH SBIR/STTR and NIH R-series. We don't do NSF, DoD, ARPA-H, or foundation grants — the rules are different, and pretending otherwise wastes your money. If you ask, we'll say no and point you elsewhere.
We'd rather lose a deal on scope than run a check that produces false confidence.
Your file is parsed in your browser. Your results are deleted after 30 days.
For Submission Match, the Aims graders, Resubmission Lab, and NOFO Rubrics, your PDF or DOCX is parsed in your browser. The file itself never leaves your machine — only the extracted text is sent to our servers to produce your result. The one exception is the Submission Readiness Check, which has to read the PDF byte layout to count pages and margins: that file is streamed to a server function, held in memory for the length of the check, and discarded. We operate no file storage bucket, so there is no PDF or DOCX archive to breach, subpoena, or leak. Nothing you upload is ever used to train models — ours or anyone else's.We will not pretend we keep nothing. To let you reopen a result we store the analysis we produce, and for Submission Match that record includes the intake narrative you entered. So we put it on a clock instead: Analysis results are auto-deleted 30 days after they are created, unless you save them to your Document Vault. A daily job performs the deletion; saved items are kept until you delete them. You can see exactly what is stored per feature, and every third party that touches it, on our privacy page — including a named sub-processor list and a per-feature data-flow table.
A larger training corpus would help our scoring accuracy, and keeping your files would make debugging a bad score far easier. We don't take either.
The rubric is published. Argue with it.
The Submission Match rubric is public: a two-track rubric — R-series and SBIR/STTR — with weighted sub-scores that sum to 100% in each track. You can read every pillar, weight, and sub-score on the methodology page before you run one. The rubric isn't the moat — the verification discipline behind it is.
Competitors can copy the rubric today. They can't copy the outcome-tracking discipline: every score we issue is being labeled with its real funding outcome as results publish, and that cohort compounds.
How to hold us to it
- Rules are stale? Email rules@grantreviewpro.com with the rule and the current NIH source. We publicly acknowledge and re-verify within 5 business days.
- Score felt wrong? Every scored application shows the exact rubric run against it. If the sub-score reads don't match what you submitted, tell us — we fix the rubric, not the score.
- Delete my data? Email privacy@grantreviewpro.com. Confirmed within 72 hours.