GrantFlagScrub

Who it's for

US academic PIs and research admins who today Ctrl-F through FOIA-leaked or news-circulated federal keyword lists in Word/Google Docs

The problem

funding — single flagged words have canceled billions in awards and derailed careers

How to build it

web upload (PDF/DOCX) + optional Google Docs sidebar

How it makes money

university research offices or individual PIs pay $19 per scan or $49/mo because general LLMs lack the exact current lists, log sensitive text, and give uncited guesses

Why it doesn't exist yet

incumbent grant platforms refuse to maintain politically radioactive blacklists; the gap is a tiny maintained scraper + redline UI an indie can ship without institutional risk

First users

authors and commenters on the Higher Ed Dive / HN thread who already keep personal keyword spreadsheets

Build size

1 person x 5 weeks — PDF/DOCX parse, keyword DB from public sources, annotated export; excludes full proposal generation or agency submission

Biggest risk

NSF/NIH publish an official scrubber portal or the lists are withdrawn from public view

Conditions for a hit (all 3 required)

• Accepts PDF or DOCX grant draft and returns annotated file with each hit highlighted
• Every highlight cites the specific public keyword list and date version used
• Produces at least one alternative phrasing per hit and finishes a 20-page doc in under 10 minutes

How it's judged (in 6 months)

Product Hunt daily top 5 or GitHub 800+ stars(judgment date 2027-02-18)

AI self-confidence 48/100 — self-reported likelihood of meeting the criterion, not a business success rate