Methodology
unwrite is not a generic chatbot. It’s an editorial desk with AI-assisted writing tied to a fixed process. It learns your voice, fits the structure to the job, grounds every claim in a source you can defend, then audits and re-reads the draft, automatically. It arrives for your review in the unwrite Studio, where every edit is yours to keep or undo.
Try unwrite freeReview, verify, and finish your piece with every edit under your control.
Hover any marker to explore the editor.
If a text sits unanswered for an hour and your whole body tenses, if a mild comment loops in your head for days, if a small dismissal lands like a physical blow, you are not the problem. For many autistic and ADHD adults, that reaction has a name: rejection sensitive dysphoria.
The difference is real, measurable, and not a defect. Brain imaging consistently shows variation between neurodivergent and neurotypical brains, according to TherapyExplained. In ADHD brains, differences in dopamine and norepinephrine signaling reshape the networks that handle attention and executive function.
The same process runs on a short email or a 40-page book/ebook, and your voice is maintained all the way to the last line.
Give unwrite a bit of your real writing. Before a single line is drafted, the Editorial Engine reads your sample to capture voice, tone, cadence, sentence rhythm, and the phrasing that makes your writing yours. The draft is written to match your style, not a predetermined tone, and your voice is held across the whole piece.
A blog post, a white paper, and an email are each built differently, so the Editorial Engine plans for each one specifically. Long-form content is drafted section by section against an outline you approve first, and the outline stays yours after drafting: reopen it, change it, and unwrite redrafts free, with your current version saved for a one-click return. Reports get an executive summary (written last) from the finished body. Web copy is tailored to the page type and the reader's position in the sales and marketing funnel. The structure fits the deliverable before any prose is written.
Give unwrite your own sources, or let it research the topic. Either way, it cites what it can verify, ranked by authority, and checks the claims back against the sources. When a claim cannot be supported by a reliable source, unwrite flags the gap rather than inventing a number. Original analysis and connective prose are unwrite's job. Your material is uploaded to unwrite, and each item is tagged by how it is used: Source, Background, Outline, or Voice. unwrite reads each item and pre-suggests its tag, always shown and changeable in one click, so what it cites as verified is exactly what you approved. Background informs the draft but is never quoted as a source.
Every draft gets a full Editorial Review, automatically: AI-pattern checks, a banned-language scan, voice continuity against your sample, and a claim-by-claim check against selected sources. You get either a clean read or a list of flagged lines, each with a suggested fix.
A fresh, independent pass runs unwrite's Second read. It reads the finished draft cold, as a skeptical editor would, marking what it would rework or cut, so opportunities for improvement can be surfaced. The Second read is included on Pro: five per piece, and two on a piece long enough to be read in parts, then 1 credit each.
Some rules are too important to leave to a model's subjective judgment. Em dashes, double hyphens, the “it's not just X, it's Y” construction, and a curated set of banned phrases are removed deterministically as prose is generated. A banned-language scan flags anything else the model missed, so it surfaces for your review before you publish. This is the last of four layers, from the model's instructions down to a mechanical pass.
Experience the unwrite Studio.
Start writing free