About StatementBench and its author
Who builds StatementBench, what qualifies him to test document extraction tools, how the site is funded, and how to reach him.
The author
StatementBench is written and maintained by one person, a Principal ML engineer, who publishes under the site's name.
- 10+ years in machine learning, from research to production
- Former Staff ML Engineer working on document extraction and OCR
- Former Lead ML Platform Engineer for a 5,000-stream computer-vision platform
- Contributor to vLLM; creator of Videoflow (1,000+ GitHub stars)
Why this site
Pulling numbers out of documents is the kind of problem I have worked on professionally: document extraction and OCR systems, and the machine learning infrastructure that runs them at scale. Bank statements and other small-business finance documents are a good place to apply that experience in public, for three reasons:
- The stakes are concrete. A wrong amount in your books becomes a wrong tax return or a reconciliation that won't close.
- Accuracy can be checked. Statements carry running balances, so extraction errors can be measured exactly, without opinion.
- The market is full of claims and short on measurements. Most converters say they are accurate. Few publish how they know.
How the site is funded
The tools are free and will stay free. The site will earn money from affiliate links to some of the vendors it covers, labeled "(paid link)" wherever they appear. The affiliate disclosure lists every current relationship and the rules that keep commissions away from results. There is no advertising and no sponsored content.
Independence
StatementBench is not affiliated with any bank, accounting software company or converter vendor. Vendors cannot pay for placement, reviews or benchmark results. The site publishes no reviews or testimonials written by anyone other than the author.
Contact
Email hello@statementbench.com to report an error, a changed price, or a statement layout the PDF tool gets wrong.
Also available as Markdown.