Completed
Closed and oversubscribed. Processing volume up 3x this year.
CursiveThis project has been verified
Handwriting-to-text OCR tuned specifically for messy clinical charts.
$600,000 raised so far, confirmed by both parties (80%)
Details
- Category
- AI / Machine Learning
- IP type
- Software
- Stage
- Scaling
- Ask type
- Fixed amount
- Listed
- Sep 16, 2026
- Views
- 1
Description
Cursive is an OCR API fine-tuned specifically on handwritten clinical notes, achieving accuracy generic OCR can't touch on the abbreviations, shorthand, and genuinely bad handwriting real charts contain. Processing over 500,000 pages a month for two regional health systems.
Problem
Generic OCR fails badly on real clinical handwriting, full of medical shorthand and abbreviations no general-purpose model has seen enough of.
Solution
A model fine-tuned specifically on labeled handwritten clinical notes, including medical abbreviation expansion, not a general handwriting OCR repurposed for healthcare.
Target market
Health systems and medical records digitization vendors.
Business model
Per-page API pricing, with an enterprise volume tier.
Investments
Kenji Watanabe invested $400,000
8% equity
Farid Karimi invested $200,000
4% equity
Discussion (1)
Farid Karimi
Medical shorthand is genuinely the hard part here, and the accuracy numbers back up the pitch.
Hana SuzukiVerified account
Innovator·🇯🇵 Japan
ML researcher focused on handwriting recognition for clinical use.