For parents

What is Alefba?

Alefba helps children aged about 7 to 14 learn to read and speak Persian: the letters, everyday words, how people really say them at home, and short practice conversations with friendly characters.

What does it keep?

On the device your child uses, and only there: the first name or nickname they type, their lessons and scores, and their settings. These are stored in the browser and never sent to us. You can clear them with Reset all progress in the Review tab.

If you sign in for the talks, Alefba's server keeps your email address, which plan your family has, and the time each talk started, so it can count talks (those times are deleted after about 100 days). It never keeps what is said in a talk. Sign-in codes are sent through Resend, an email delivery service. Before a code is sent, a quick robot check from Cloudflare (called Turnstile) makes sure a person is asking, so nobody can use the site to send lots of emails.

Payments are handled by Stripe, so Alefba never sees your card number. One cookie keeps you signed in on that device. Alefba has no advertising, no analytics and no tracking cookies.

To delete your account, open the Review tab, then Account and plans, then Delete account, and enter the code we email you. That ends any plan, removes your email address and your family's talk counts from Alefba, and deletes your customer record at Stripe (Stripe keeps its own records of past payments). The only thing Alefba keeps is a scrambled code made from your email address, used just to remember that the free talks were used, so they aren't given again.

How do the AI talks work?

In a talk, your child chats with a character, like a new classmate or a grandmother, using the Persian words they've learned. The character's replies are written by Claude, an AI made by Anthropic. Every talk is labeled AI on screen, and the characters never pretend to be real people.

• Talks start only after a parent signs in with their email on that device, using a code we send. Signing out turns talks off again.
• What your child types in a talk is sent through the Alefba server to Claude so it can write the reply. The Alefba server doesn't save the chats.
• Your child's name is never sent, and neither is your email address. Email addresses, phone numbers and links your child types are removed before anything is sent.
• Anthropic handles the messages under its commercial terms, which don't allow it to train its AI on them. It deletes them within 30 days, except in limited cases, such as a message flagged for a safety review. See how long Anthropic keeps this data.
• To stop the service being overused, the server briefly keeps the internet address a request came from, in memory only, for up to an hour. To limit sign-in emails, it also counts them under a scrambled code made from that address, deleted within three days.

What do the plans cost?

The lessons are free. Every family gets 3 free talks, and after that:

Plans renew each month until you cancel. To cancel, switch plans or update your card, open the Review tab, then Account and plans. When you cancel, your plan stays until the end of the month you've paid for. Plans and payments are for grown-ups, so the app asks a quick sum before showing them, and switching plans or deleting the account also needs a code we email to you.

The rules, in short

• A parent or guardian signs in and looks after the account. Children use Alefba with their permission.
• Plans renew each month until you cancel. If prices ever change, your plan keeps the price you signed up for.
• Refunds: if something isn't right, write to us within 14 days of a payment and we'll refund it.
• Unlimited means up to 30 talks a day, to keep things fair for everyone. We may pause an account that misuses the talks, for example by trying to make the characters say harmful things.

How are the talks kept safe?

Every character is told to keep everything suitable for young children, to steer kindly away from anything upsetting, and never to ask for personal details such as an address, school or phone number. Children are also reminded on screen not to share those things.

AI can still make mistakes. Like anything online, it's a good idea to look at a talk together now and then and to remind your child never to share personal information.

If a talk ever shows something that worries you, please tell us (see below) so we can fix it. Chats aren't saved, so a screenshot or a short description helps.