Your grant section, scored by the reviewer before you submit it.
Paste an SBIR, NIH, or NSF solicitation. We extract the actual review criteria, draft your sections against them, then run a hostile reviewer simulation that scores each section 1-9 and tells you exactly what will get you triaged. $49 per section. $149 for the full pack.
1. Paste your solicitation
Paste the relevant portions: review criteria, eligibility, required sections, page limits, formatting rules. We don't need the whole 80-page document — just the parts that define how you'll be scored. This step is free.
Your draft workspace
Payment confirmed. Draft each section, then run the reviewer scorecard. Drafts and scores are saved in your browser — close this tab and come back later if needed.
Why this exists
Hiring a grant writer for an SBIR Phase I costs $3,000–$15,000. ChatGPT will write you a section for free, but it won't tell you whether the section will actually score well — and it doesn't know what review criteria your specific solicitation uses.
GrantForge does both. It writes against the actual published criteria, then plays the role of a tired reviewer with 15 proposals on their desk who is looking for reasons to mark you down. The scorecard is the real product. The draft is the warm-up.
What you'll get
- Compliance checklist (free) — every page limit, font rule, mandatory certification, and formatting requirement extracted from your solicitation.
- Section draft (paid) — written in agency house style, with placeholders marked clearly where you need to insert specific data the AI shouldn't fabricate.
- Reviewer scorecard (paid) — 1-9 NIH or 1-5 NSF/DoD score per criterion, fatal flaws flagged separately from major and minor concerns, and the top 3 specific edits that will lift your score the most.
What this is not
This is a drafting and review aid, not a submission service. The AI does not verify scientific accuracy, does not check your citations, and does not guarantee funding. Federal review is human, subjective, and partly luck. Use the output as a strong first draft and a private peer review — then have a human collaborator read it before you submit.
FAQ
Which agencies are supported?
Anything that publishes review criteria in its solicitation: NSF SBIR/STTR, NIH R-series and SBIR, NIST SBIR, DoD SBIR/STTR, DoE SBIR, USDA SBIR, NASA SBIR. The parser reads whatever you paste — agency-specific quirks (NSF's Project Pitch, NIH's Specific Aims one-pager) are recognized automatically.
Can I get a refund if the output is bad?
Yes. Email refunds@[your-domain] within 7 days with the session ID and a one-line note about what didn't work, and we'll refund you. We'd rather know what's wrong than fight a chargeback.
Is my project idea private?
Your text is sent to Anthropic's Claude API, governed by their data policy (no training on inputs by default). We don't store your project description on our servers. We log payment events (Stripe session ID, tier, email) for accounting only.
What if my solicitation is custom or non-SBIR?
It still works as long as the solicitation publishes review criteria. Foundation grants, state-level innovation funds, and corporate research grants all work. The output adapts to whatever criteria you paste.