Clear writing is rewriting. This tool takes dense text and rewrites it at the reading level you choose, using your own OpenAI or Anthropic key, then estimates the Flesch-Kincaid grade before and after so you can see the change — all in your browser.
How it works
Choose a provider and model, paste your API key, drop in your text, and pick a target grade level from very simple (grade 5) to specialist (grade 13+). The tool estimates the current reading grade locally, then sends one direct request asking the model to rewrite at your target level using shorter sentences and simpler words while preserving meaning, facts, and any keywords you list. It estimates the new grade on the result so you can compare.
For Anthropic, the request includes the official direct-browser-access header so it works straight from the page.
What Flesch-Kincaid measures
The grade level is driven by two things: how long your sentences are and how many syllables your words have. Shortening sentences and swapping long words for short ones lowers the grade — which is exactly what the rewrite targets.
Tips
- Aim for grade 7-9 for general audiences; most successful web copy sits there.
- Use the keyword field to protect terms that must stay exact — drug names, legal terms, product names.
- The local grade estimate is a guide, not gospel; read the rewrite aloud as the real test of clarity.
When to use a readability rewriter
This tool is most useful in three situations. First, when you have written something correct but dense — technical documentation, policy text, or an email written in a rush — and you need it to land with a general audience. Second, when you are adapting content written for specialists to reach a broader readership: a clinical guideline to a patient leaflet, a legal clause to a plain-English summary, an engineering spec to a product description. Third, when you are testing whether your own draft is clearer than it feels — comparing the before and after grade is a fast diagnostic.
It is less suited to short, punchy copy that already reads well, or to content where every word has been carefully chosen for legal or contractual precision. Rewriting a contract clause for clarity can inadvertently change its meaning; proof any such rewrite carefully against the original intent.
How Flesch-Kincaid actually works
The Flesch-Kincaid Grade Level formula combines two inputs: average sentence length (total words divided by total sentences) and average syllable count per word (total syllables divided by total words). A long sentence with long words pushes the grade up; short sentences with short words bring it down. The formula does not measure clarity directly — it measures the surface features that correlate with readability difficulty.
This matters because it means the score can be gamed. You can produce a low grade score with short, choppy sentences and simple words while still writing confusingly if the ideas are poorly organised or the logic jumps around. Use the grade as a directional check on vocabulary and sentence structure, then read the text aloud as a check on whether it actually flows.
For most public-facing web content, a grade 7–9 result lands with the broadest audience while still reading like a competent adult wrote it. Grade 5–6 suits patient information, instructions for products used in stressful situations, or content for audiences with lower average literacy. Grade 10+ is appropriate for specialist audiences who expect technical depth.
What the keyword field protects
When you list terms in the keyword field — for example “paracetamol”, “Section 21 notice”, or “gross domestic product” — the model is instructed to keep those terms unchanged even if it would normally simplify them. This preserves precision where it matters: a medical leaflet might target grade 6 overall but must use the exact drug name; a legal summary might simplify every clause except defined terms that carry specific legal weight. For a worked example, paste a paragraph of clinical or legal text, set target grade 7, and list your protected terms — the rewrite will simplify the structure without touching the terminology you locked.