Italian writing practice with AI feedback
The written-interaction paper is worth 35 of the 100 points in the Italian A2 test. Here you practise replying to e-mails and invitations, filling in forms, and writing letters and complaints, with instant AI feedback. Free, in your browser.
Each task is a different text type — a reply to an e-mail, a form to fill in, a formal letter, a complaint or a small ad each has its own rules, so the AI grades your text against the rules of its form. Every task’s type is shown on its badge.
Pick a task and start — the AI grades what you write:
Rispondere a un’e-mail (test A2)
Reply to an e-mail (A2 test task)Written interaction — reply🗂️Compilare un modulo (test A2)
Filling in a form (A2 test task)Written interaction — form
Lettera a un amico
Letter to a friendInformal letter
Lettera formale — una richiesta
Formal letter — a requestFormal letter
Reclamo a un negozio
Complaint to a shopComplaint
Imparare l’italiano — testo di opinione
Learning Italian — opinion pieceOpinion piece
Recensione di un film
Film reviewReview
Relazione su una gita
Report on a day tripReport
Un racconto
A short storyStory
La mia città in Italia
The town where I liveDescription
Invito a una festa
Party invitationInvitation
Annuncio — stanza in affitto
Small ad — room to rentSmall ad
Cartolina dalle vacanze
Holiday postcardPostcard
Il telefono e la vita di ogni giorno
Phones and everyday lifeOpinion piece
E-mail a un collega
E-mail to a colleagueSemi-formal e-mailWhy writing is the part that decides the A2 test
In the Italian A2 state language test the written-interaction paper is a single task, but it is worth 35 points out of 100 — more than any individual question in the other two papers. You are asked either to reply to an e-mail, a postcard or an invitation supplying, as a rule, three specific pieces of information, or to fill in a form: enrolling on a course, personal details, a residence-permit application, a request for a benefit.
The score is not counted error by error. It is a banded judgement on task completion: up to 35 points if you supplied every piece of information asked for (or filled in every field of the form), up to 28 if the answer is only partial, 0 if it cannot be assessed. The practical consequence follows: answering two of the three questions costs far more than a handful of grammar mistakes.
And there is no word limit at all, neither minimum nor maximum. A three-line message containing all three pieces of information is a complete message. Writing at length to "fill the space" does not help; covering everything does.
How the AI practice works
Write your answer in the box and send it. The AI gives you a score out of 15 across three criteria, a list of what was missing against the prompt, the mistakes worth fixing, and an example of a good answer to compare with your own. The tasks that reproduce the test paper use the same banded criteria as the real one.
Alongside the two test-format tasks you get the written forms real life in Italy demands: the formal letter to the Comune, the complaint to a shop, the small ad to rent out a room, the e-mail to a colleague, the postcard. Every task has a "Rules" panel with the expected structure and useful phrases, and a "Help" panel with the key words translated into your language.
All free: before an AI evaluation you are only asked to confirm you are not a bot.
Frequently asked questions — writing
How many words do I have to write in the written-interaction paper?
No set number: the test specification sets neither a minimum nor a maximum word count. What counts is having answered every piece of information asked for — usually three — or having filled in every field of the form.
What exactly am I asked to do?
One of two task types: reply to an e-mail, a postcard or an invitation supplying the information the prompt asks for, or fill in a form (course enrolment, personal details, residence-permit applications, benefits, a bank account).
Will grammar mistakes fail me?
Not on their own. The criterion is task completion, and the specification itself asks assessors to value communicative effectiveness over formal accuracy. A text with a few errors but complete is worth more than a correct text that is missing a piece of information.
How is the score calculated in this practice?
Your text gets a score out of 15 across three criteria, with a comment on each and an example of a good answer. On the test-format tasks that scale corresponds proportionally to the official bands of the real paper.
Do I have to pay?
No. The practice and the AI feedback are free; before submitting you are only asked to confirm that you are not a bot.