Which tool to choose
You already have the text and want a clean, readable rewrite: use a conversational assistant, not an automatic "paraphraser" (the ones born to fool AI detectors are of little use here).
For a long text or an entire document → ChatGPT with the Canvas, the editing space alongside the chat. There you rewrite individual paragraphs, reorganize sections, and compare versions: you see the before and the after and fix only the weak points. The free version is enough to rephrase and summarize.
If you work inside Google Docs, Gmail, or Slides → Gemini, integrated into the Google ecosystem. It helps you write, rephrase, and summarize without leaving the document you're already working on. It's not specialized in persuasive writing, but it makes communications clear and consistent with the tools at hand.
If the naturalness of the final writing matters → Claude. It tends to be less "moralistic" than GPT models and uses more varied sentence structures by default.
For most cases, the free version of any of the three is enough. The prompt below works identically on all three.
How to do it
From a browser or from the app the path is the same: open a new conversation, paste the prompt, then below it paste your text.
- Open the chosen tool and create a new chat.
- Decide first what you need: only inclusive (gender), only accessible (readability), or both. Change the prompt accordingly.
- Copy the operational syntax below and paste your text in the indicated spot.
- Read the output. Don't stop at the first version: ask for specific changes ("sentence X is still too long," "here you used the schwa, remove it").
- Reread it yourself, out loud. If you stumble, the reader stumbles too.
The operational syntax for the dual goal, inclusive and accessible:
Rewrite the text you find at the bottom following these rules.
Gender-inclusive language:
- use the correct feminine forms for roles and professions (engineer, lawyer, director, minister)
- instead of the generic masculine, prefer collective forms (the staff, the clientele, those who participate) or the double form (workers, both women and men)
- do not use the schwa, asterisks, or @ signs: they hinder screen readers and people with dyslexia
Accessibility and readability:
- one sentence, one concept: break up long periods
- use the active voice and subject-verb-object order
- choose words in common use instead of technical terms, and when a technical term is necessary, explain it in parentheses at its first occurrence
- avoid double negatives
- keep the content and the data unchanged: do not add and do not remove information
At the end, list in three lines the main changes you made.
Text to rewrite:
[paste here]
After the output, check that numbers, dates, and proper nouns have stayed identical to the original. It's the point where the AI most often makes mistakes.
These rules are not an invention. "Plain language" calls for short sentences, in active form and without subordinate clauses: the information stays the same, but the text becomes accessible to those with a cognitive disability, a learning difficulty, or who struggle to process information.
A concrete example
Let's start from a real corporate email, written by an HR manager:
"It is hereby communicated to all users that, should they not have yet proceeded with the completion of the registration procedure, they are obligated to comply by no later than the peremptory deadline indicated above, under penalty of exclusion."
Pasted into the prompt above, ChatGPT in the free version returned:
"We're writing because we don't yet see your registration. Please complete it by the date indicated above. After that date it will no longer be possible to register."
What changed: the generic masculine "all users" gives way to the direct "you," which includes anyone reading; the three subordinate clauses become three single sentences; "completion of the registration procedure" becomes "complete it"; "under penalty of exclusion" becomes an explained consequence. The meaning is identical, the date stays the date. A text that previously repelled is now understood on the first read, even for those with cognitive difficulties or learning the language.
When it does NOT work (and how to fix it)
If the AI changes the numbers or the dates
It's the most serious risk: the AI rephrases and in doing so "rounds" a figure or shifts a deadline. Fix: in the prompt write "do not modify numbers, dates, proper nouns, and amounts," and after each rewrite compare the data line by line with the original. For the texts you really count on, keep the original open alongside and reread only the figures.
If it slips in schwas, asterisks, or dashes
It happens because many models associate them with inclusiveness. The Accademia della Crusca, responding to a request from the Court of Cassation, has advised against graphic signs like the asterisk and the schwa: they do not conform to spoken language and create inconsistency, especially in legal language. Fix: reply "remove every schwa and asterisk, use collective forms or the double form." The alternatives cover almost every case: the double form ("students, both women and men"), collective nouns ("the teaching staff"), neutral forms ("whoever is interested").
If the text stays complicated
Sometimes the AI "simplifies" only on the surface and leaves long periods. Fix: ask for an explicit level ("rewrite at a middle-school reading level, maximum one subordinate clause per sentence"). For an objective check use READ-IT, the readability analysis tool for Italian developed by the CNR: it assesses word and sentence length, vocabulary, and syntax, and tells you how understandable the text is instead of relying on your eye.
If it distorts your tone
In rewriting, the AI can make everything flat and impersonal. Fix: give it an example of your style ("keep a warm and direct tone, like this sentence: …") and have it rewrite only the problematic points, not the whole text. ChatGPT's Canvas is for exactly this: you work on a single paragraph without touching the rest.
A tip from someone who really uses it
Inclusiveness isn't a list of rules to apply mechanically. The typical mistake is treating inclusive language as a set of fixed, universal rules: a journalistic text, an advertising one, and a narrative one call for different adaptations. Use the AI for the first heavy pass, but the last word is yours: you know who will read and in what context. And remember the most underrated principle of accessibility: easy-to-read texts are the ones that breathe, the opposite of walls of text. After rewriting the sentences, divide the text into short blocks, with white space and clear subheadings.
Frequently asked questions
Can I rewrite long texts all at once?
It depends on the tool's limits. Dedicated paraphrasers have low caps: Neural Writer, for example, is free and registration-free, but doesn't accept texts over 10,000 characters. With ChatGPT, Claude, or Gemini you handle much longer documents, but it's better to proceed in blocks: on huge texts the quality drops and you lose control over the data. Working in sections also lets you correct each piece right away.
Do I absolutely have to pay?
No. To rewrite a text, the free version of ChatGPT, Gemini, or Claude is enough. The paid plans serve high volumes or advanced features: ChatGPT's Go plan costs about €4 a month and Plus about €20 a month (VAT included). For occasional rewriting you don't need them.
Are "inclusive" and "accessible" the same thing?
No, and confusing them leads to errors. Inclusive concerns above all gender and respect for people (gendered forms, collective forms). Accessible concerns understandability for those with cognitive disabilities, dyslexia, or a low level of education. Sometimes they're in tension: some "inclusive" signs worsen accessibility, because the excessive use of schwas, asterisks, and dashes slows reading and creates difficulty for those with dyslexia or who use assistive technologies. That's why in the prompt you ask for both things together: they balance each other.
Is inclusive language just an ideological fad?
It's the most common objection, and it deserves a serious answer. The feminine forms we're talking about aren't recent inventions: engineer, doctor, lawyer, minister, notary are correct forms already attested in Italian grammar. And they don't come from a passing fad, but from over thirty years of institutional work: public administrations have adopted gender-attentive language since guidelines in the 1980s. You can choose how far to go, but writing "the director" (feminine) instead of "the director" (masculine) when the role is held by a woman is correct Italian, not militancy.