Your detector report already told you which paragraphs to touch
Most students do not need a full rewrite. They need a careful pass on the spans Turnitin or iThenticate marked — without breaking citations, formulas, tables, or the numbers they already verified.
Angle 1 — How to read an AI detection report without panicking
An AI writing score is not a plagiarism verdict and it is not a moral judgment. It is a model’s estimate that some stretches of your document look statistically similar to machine-generated prose. Different tools disagree. The same paragraph can look “clean” in one viewer and “flagged” in another. That disagreement is why a blanket rewrite of the whole draft is often the wrong move: you spend hours changing prose that was never the problem, and you risk introducing new inconsistencies in quotes, page numbers, and method details.
Start with the report as a map. Open the highlighted spans. Ask three plain questions for each one: (1) Is this my voice or a pasted paraphrase? (2) Does it contain claims that must stay numerically exact? (3) Would a human reader in my field find the wording oddly generic? If the answer to (1) and (3) is “generic paste” and (2) is “no critical numbers,” that span is a candidate for rewrite. If the span is a methods table, a quotation, or a formula block, leave it alone and document why.
HumanPen is built for that workflow. You import the detector report (or select paragraphs yourself), and the tool rewrites only the flagged stretches. Layout, references, and numbers are treated as constraints, not decoration. Billing follows actual rewritten words — not a flat “whole document” fee — and new accounts receive 100 free credits ≈ the first 1,000 words so you can test the loop on a real chapter before paying.
Angle 2 — What “document-level rewrite” means when a second check still shows ≥20%
Product pages sometimes say “we rewrite until under 20%.” Read that as a process description, not a guarantee that any specific journal, school, or vendor will accept your submission. Detectors change. Institutional policies change. The honest promise is narrower: after you run HumanPen on flagged spans, if a recheck of the same report still shows ≥20% on remaining flagged paragraphs, those still-flagged spans can be re-run without an extra charge under the product’s stated free-retry rule. That is Condition C style document-level rewriting: iterate on the residue, not on marketing slogans.
Why does that matter for students and thesis writers? Because the expensive failure mode is random paraphrasing. Random paraphrasing flips synonyms, breaks idiom, and sometimes invents softer claims that no longer match your data. A document-level pass keeps the argument spine intact: same section order, same citations, same numbers, different surface prose where the detector pointed. You stay accountable for the ideas; the tool helps with the flagged texture.
Practical sequence we recommend: upload or paste the chapter → attach or mark the AI-report highlights → rewrite flagged blocks → export → recheck with your institution’s tool → if residue remains ≥20% on those blocks, use the free re-run on the leftover flags → stop when the report is under the threshold you are targeting or when further changes would damage meaning. Then do a human proofread for voice. HumanPen does not replace your judgment; it shortens the loop between “report says here” and “prose changed here.”
What HumanPen is — and is not
- Is: a document-level AI rewrite helper aimed at detector-highlighted paragraphs, with credit-based billing and a free starter allotment (~1,000 words).
- Is not: a promise that you will “pass” any named detector, a ghostwriting service for original ideas, or a substitute for citation ethics.
- Is not: “forever free for the whole site.” Free credits are for first trials; continued use is metered by rewritten words.
If you are choosing a tool this week, prefer ones that (a) show you exactly which spans will change, (b) preserve references and figures, and (c) price by rewritten words. That combination keeps cost proportional to the problem the report actually showed you.
A short checklist before you spend credits
Before you burn free or paid credits, freeze a copy of the chapter and the original detector PDF. Note the tool name, the date, and the overall percentage. Then list every highlighted span with a one-line reason: “generic methods prose,” “intro filler,” or “keep — quotation / table.” That list becomes your rewrite budget. HumanPen’s free allotment — 100 credits ≈ 1,000 words — is enough for a first pass on a typical flagged cluster, not for inventing a new thesis overnight.
When you export, skim the changed spans side by side with the report. Confirm that citations still resolve, that numbers match your tables, and that you did not soften a claim your advisor already approved. If a second check still shows ≥20% on leftover flags, use the product’s free re-run on those residues rather than starting from a blank page. That is the honest loop: report → targeted rewrite → recheck → residue only.
Finally, remember institutional context. Some departments ban AI writing tools entirely; others allow editing aids with disclosure. HumanPen is an editing aid aimed at detector-highlighted texture. You remain responsible for originality, citations, and whatever disclosure your school requires. Prefer clarity over hype: try the free credits on a real flagged section, judge the output yourself, then decide whether paid credits are worth it for the rest of the manuscript.