Three intensity modes (Light, Balanced, Maximum). One-click rewrite on only the flagged sentences. Built-in detector re-scores after every rewrite so you can watch your Authenticity Score climb.
3 free scans/day · Light + Balanced modes included · Maximum mode on Starter ($9.99/mo)
Paste your draft, choose Light / Balanced / Maximum, and let the AI rewriter rewrite the flagged sentences. The built-in detector re-scores instantly so you can verify the rewrite landed.
Open the AI Rewriter →Minimal changes. Preserves your voice and sentence structure. Touches only the most obviously-AI phrases.
Best for: personal writing, blog posts where you want your voice intact.
A fuller rewrite of the flagged sentences while keeping the document structure. The default for most drafts.
Best for: articles, marketing copy, mid-stakes essays.
Aggressive rewrite tuned for the lowest possible AI score. Bigger phrasing changes; structure may shift.
Best for: academic submissions, content audits, compliance scenarios where the score matters more than the voice.
Paste your draft to get an AI score and sentence-level highlights — the AI rewriter only needs to rewrite the flagged sentences, not the whole text.
Click Light, Balanced, or Maximum. The AI rewriter rewrites only the flagged sentences, preserving citations, numbers, and document structure.
The detector re-scores your rewritten text immediately so you can see your Authenticity Score climb. Re-run with a higher mode if needed.
Rewrite AI-drafted study material in your own voice before submission. Balanced mode keeps the argument structure but rewrites the surface phrasing.
Take a fast AI first draft and rewrite it before publishing. The Authenticity Score gives you a quantitative target instead of guessing whether "it reads right."
Bulk-rewrite client deliverables. API access on the Business plan ($29.99/mo) covers 10,000 calls/mo for production workflows.
Rewrite AI-summaries to match your departmental tone. Light mode is conservative enough to preserve technical terminology.
It's not a plagiarism shield. The AI rewriter rewrites the surface language; it doesn't change the underlying source of your ideas. If your content was lifted, the AI rewriter won't fix that — use the plagiarism checker for that workflow.
It's not a guaranteed pass on every detector. No AI rewriter guarantees a pass on any specific third-party detector. The Authenticity Score we report is computed against our own detector, which we tune to align with general industry signals. If you need to pass a specific detector (e.g. Turnitin), verify by re-scanning on that tool.
It works best in English. The AI rewriter is English-trained. It runs on other Latin-alphabet languages but quality drops outside English.
Paste your text and pick an intensity mode — Light, Balanced, or Maximum. The AI rewriter rewrites the AI-flagged sentences while preserving your meaning, then the built-in detector re-scores the result so you can see your Authenticity Score climb. Most rewrites take under five seconds.
Light: minimal changes, preserves your voice — best for posts that just need a polish. Balanced: a fuller rewrite for the flagged sentences while keeping structure — best for blog drafts and articles. Maximum: aggressive rewrite for content that absolutely must pass detection — best for academic or compliance use cases.
Yes. The AI rewriter preserves the semantic content — claims, facts, structure, citations — and rewrites the surface phrasing. If a sentence cites a number or attribution, those are preserved. Light and Balanced modes are conservative; Maximum mode is more aggressive but still factually faithful.
The detector + AI Rewriter combo is free on 3 scans per day with Light and Balanced modes. Maximum mode is on the Starter plan ($9.99/mo) and above. There's no separate per-rewrite cost — one scan covers both detection and rewriting.
A paraphraser rewords text without checking AI signals. An AI rewriter is targeted: it rewrites the specific sentences flagged by an AI detector, using transformations designed to drop the AI-probability score. TextSight ships both — see /tools/paraphraser/ for the general paraphrasing tool.
Yes. The AI rewriter rewrites the language patterns that detectors flag — those patterns are model-agnostic for the most part (burstiness, perplexity profiles, lexical fingerprints). Output from GPT-4, Claude, Gemini, and Llama 3 all respond to the same rewrite passes.
It typically scores lower on most detectors after a Balanced or Maximum rewrite — but no AI rewriter guarantees a pass on any specific third-party detector. The Authenticity Score we report is computed against our own detector, which we tune to align with general industry signals. If you need to pass a specific detector (e.g. Turnitin), verify by re-scanning on that tool.
Yes — the AI Rewriter page in the app (/ai) lets you paste text and rewrite directly. But running the detector first is almost always more useful, because you only need to rewrite the flagged sentences, not the whole draft.
Six concrete rewrite passes the AI rewriter applies to flagged sentences. Every pass is deterministic, reversible if you don't like the output, and runs in under five seconds for a typical 500-word draft.
AI drafts tend to produce sentences of similar length, which detectors call low burstiness. The AI rewriter rewrites flagged blocks so short, punchy sentences sit next to longer, more discursive ones, mirroring how real writers pace a paragraph.
Words like utilize, facilitate, leverage, demonstrate, and ascertain are statistical fingerprints of model output. The AI rewriter swaps them for the plain-English equivalent (utilize becomes use, facilitate becomes help, demonstrate becomes show) without changing what the sentence means.
Models love rhythmic triplets and identical clause patterns (the system is fast, the system is reliable, the system is secure). The AI rewriter rewrites two of the three clauses in different grammatical shapes so the prose stops sounding rehearsed.
It is important to note that, in many cases, generally speaking, it should be noted that. These hedges add zero meaning and signal AI drafting. The AI rewriter cuts them and tightens the clause that follows, giving the sentence a clearer point.
Most AI prose avoids contractions, which reads stiff. The AI rewriter converts do not to don't, it is to it's, will not to won't where the surrounding tone allows. Academic blocks (abstracts, citations) keep the formal register; conversational blocks loosen up.
Numbers, named entities, quoted text, and citation markers (Smith 2024), [1], or footnote anchors are detected and locked. The rewrite passes flow around them. Your facts stay intact; only the connective phrasing changes.
Four tiers, picked by how much text you rewrite per week and whether you need Maximum mode or API access. Most single-writer accounts land on Pro.
Best for: Writers rewriting one or two pieces a week. Students rewriting a single essay before submission. Anyone testing the rewrite quality before committing to a paid plan.
5,000 AI rewriter characters per day, 10,000-character lifetime cap before signup is required, Light and Balanced modes included.
Best for: Active students rewriting 3 to 5 essays per week. Casual bloggers rewriting 1 to 3 posts weekly. The first plan that unlocks Maximum mode for high-stakes submissions.
20,000 AI rewriter words per month, Maximum mode unlocked, Chrome extension, plagiarism risk indicator.
Best for: Daily newsletter writers, SEO contributors handling 10 plus pieces per week, freelancers running client drafts through the AI rewriter. The sweet spot for single-seat power users.
50,000 AI rewriter words per month, unlimited detector scans, file plus URL upload, priority support, all three intensity modes.
Best for: SEO agencies rewriting client deliverables at scale. EdTech and SaaS teams piping authenticity into a pipeline through the API. Editorial teams with 3 plus writers.
100,000 AI rewriter words per month, REST API access with SSE streaming, 5 team seats, white-label PDF reports.
Annual billing saves 25 percent, dropping Pro to $14.99/mo and Business to $29.99/mo. Full pricing →
Yes. Claims, facts, named entities, numbers, and citations are detected and locked before the rewrite begins. The AI rewriter changes phrasing, sentence length, and connective words. It does not change the substance. If a flagged sentence says revenue rose 12 percent in Q3 2024, the rewrite keeps the 12 percent and the Q3 2024. Always read the output before publishing; the AI rewriter ships a draft, not a final cut.
Standard (covering Light and Balanced) applies hedge removal, Latin-verb replacement, and contraction injection while keeping your sentence order and structure largely intact. It is the default for blog posts, articles, and most marketing copy. Maximum runs every pass plus aggressive sentence reordering and parallel-structure breaking. Use Maximum when the AI score has to drop below a hard threshold (academic submissions, compliance reports, content audits). Standard preserves your voice better; Maximum produces the lowest scores.
Three common reasons. First, the rewrite was on Light mode and the source text was deep into the Likely AI band; you need Balanced or Maximum to move the needle. Second, the text contains long passages of formal academic register (definitions, structured arguments) that legitimately resemble AI prose; even human PhD writing scores 30 to 50 percent AI on most detectors. Third, you are scoring against a different third-party detector whose calibration differs from ours. Run the result through our built-in detector to see your TextSight Authenticity Score, then escalate to Maximum if the score is still below your target.
Two passes are fine and often help when you escalate from Balanced to Maximum. Three or more passes risk semantic drift: each pass rewrites against the previous output, and small phrasing shifts compound into meaning changes by the fourth iteration. The recommended workflow is detect, rewrite once with the right mode, verify the score, and only re-run if you change the mode. If you are still flagged after two passes, edit the worst sentences manually instead of looping.
A paraphraser rewords text without checking AI signals. It will happily produce another AI-sounding sentence because it has no idea what triggered the flag in the first place. An AI rewriter is targeted: it reads which sentences the detector flagged, identifies which patterns (burstiness, lexical fingerprints, parallel structure) caused the flag, and rewrites only those signals. TextSight ships both. Use the paraphraser when you want a fresh wording for a single sentence. Use the AI rewriter when you want a draft to stop scoring as AI. See /tools/paraphraser/ for the standalone tool.
Citation markers are detected during pre-processing and pinned. The AI rewriter recognises APA in-text format (Smith, 2024), numbered footnote anchors [1], [2], DOI strings, bracketed URLs, and parenthetical author lists. Surrounding prose is rewritten around them so the citation lands in the same logical position in the new sentence. Block quotes are not rewritten at all; only the framing sentences are. If you use a non-standard citation style, paste a sample to support so we can extend pattern recognition.
Approximately. The AI rewriter targets a word-count delta within plus or minus 8 percent on Light mode, plus or minus 12 percent on Balanced, and plus or minus 18 percent on Maximum. Hedge removal usually shrinks the draft; sentence-length variation can lengthen specific blocks. If you need a hard length target (essay word counts, ad copy character limits), use Light or Balanced and check the output word count in the editor before submitting. The Word Counter tool at /tools/word-counter/ gives instant feedback.
English is the primary supported language and quality is highest there. The AI rewriter also runs on Spanish, French, German, Portuguese, Italian, and Dutch with respectable but lower-quality results (most rewrite passes are tuned on English lexical patterns). For non-Latin scripts (Hindi, Arabic, Chinese, Japanese), the AI rewriter falls back to a conservative paraphrase mode rather than the full pass stack. Hindi support is roadmapped for the next major release.
Detect, rewrite, verify — one tool, three intensity modes. 3 free scans/day.
Targeted rewrite guidance for the specific kind of draft you are polishing.
Practical guides for editing AI drafts into clear, natural, own-voice writing.