Skip to content

How it works

There is no model to guess with. A fixed pipeline of readable rules processes your text step by step, so the same input and settings always give the same result.

  1. 1

    Paste your text

    Type or paste into the workspace, or import a TXT or DOCX file.

  2. 2

    Pick a style and intensity

    Choose how formal, concise or conversational the result should be, and how much the engine may change.

  3. 3

    Protected text is set aside

    URLs, quotations, numbers, citations, names and your own protected phrases are hidden from the rules, then restored exactly.

  4. 4

    Rules are applied sentence by sentence

    Phrase and vocabulary rules only fire when the surrounding words support them. Sentence templates reorder clauses only where that is grammatically safe.

  5. 5

    Every rewritten sentence is checked

    A validator compares each sentence with the original. If numbers, names or structure look wrong, the original sentence is kept.

  6. 6

    You review the result

    See an honest summary of the changes actually made, compare side by side, then copy, download or accept.

The technical pipeline
  1. Validate input (encoding, length, plan limits)
  2. Normalise whitespace without damaging formatting
  3. Detect and mask protected spans
  4. Split into paragraphs, sentences and tokens
  5. Apply context-specific phrase rules
  6. Apply safe vocabulary alternatives
  7. Apply supported sentence transformations
  8. Validate each sentence against the original
  9. Compute the word-level difference
  10. Restore protected spans
  11. Return the result with a transparent change summary