Why is it important

Can't AI just do this?

29,424

patient documents checked for reading level. Most sat above the level recommended for them.

That has not improved since 2001.

Awareness was never the problem. Rewriting is.

Three models were asked to simplify the same materials. They landed at 7.6, 6.6 and 5.6. None of them the target.

A single pass has no way to check itself. Ask for grade 6 and you get "simpler," which is not the same thing.

Bar chart of mean grade-level readability score by AI model (ChatGPT, Gemini, Claude) against the original text, across three health organization sites
Will et al., JMIR 2025;27:e69955

So we made it measure itself.

Every rewrite gets scored and shown next to the grade you asked for, including when it misses.

Try LevelUp Reader

Okuhara et al., Patient Education and Counseling, 2025. Will et al., JMIR, 2025.