Say the right thing.
First time.

An AI reply assistant for any message. Give it what someone sent you — paste it, screenshot it, or use the SayThis keyboard without leaving the app — choose how you want to sound, and it writes the reply. You read it and you send it.

The ones you reread four times before answering. A friend cancelling again, a message from a date you do not want a second one with, a manager adding work on a Friday, a stranger offering half on a marketplace listing, a family group chat you cannot afford to say yes to. It works on texts, DMs, email and group chats.

Because a phone keyboard genuinely cannot read your conversation. On iOS the system returns zero characters of the thread to a keyboard extension, and on Android a messaging thread lives in a list the keyboard cannot see. Rather than pretend otherwise, SayThis reads the screenshot you take — one gesture you already know, and it captures the whole thread. It reads it on the phone, so for most screenshots no image is uploaded at all. You can also paste the message or type the point you want to make.

No. It never sends a message on your behalf, and it never will. SayThis writes the draft and puts it in your message box; reading it and pressing send is always yours.

Seventeen, and the one that matters most is Auto — the default, which reads the message and picks the voice it needs. The other sixteen are there to overrule it: Natural, Friendly, Confident, Professional and Casual are the ones most people reach for, and behind More Styles there are Funny, Flirty, Empathetic, Firm, Apologetic, Persuasive, Calm, Supportive, Direct, Warm and Enthusiastic. One at a time — tapping a tone selects it and drops whatever was there.

Yes, without going back a screen. Shorter keeps the point and loses the words. Longer adds the context the conversation already supports. Another writes a genuinely different reply rather than the same one reworded. Change tone rewrites what it says in a different voice.

Yes. SayThis replies in whatever language the conversation is in — it works that out from their messages rather than translating through English — and you can pin a language in Settings if you would rather always answer in one.

It reads them properly: who said what, by name where the app shows a name, so a reply meant for Priya does not answer Marcus. It only uses names that actually appear in the conversation.

Yes — that is what the optional second field is for. Tell SayThis the point you want to get across, in your own words, and it writes the message that lands it. It never quotes your note back.

They are not kept, and no setting keeps them. Most screenshots are never uploaded in the first place — your phone reads them and sends the text. A conversation is used to write the reply and then dropped; an image that does go up is deleted the moment it has been read. Asking for a shorter or different version sends the conversation again from your phone, because there is no copy of it on our side to reuse. Analytics record that a reply happened, never what it said.

It is the thing to think about before you paste, and SayThis says so before it reads anything. What you hand it is usually somebody else’s writing, and they have not agreed to any of this — a screenshot of a group chat sends everything visible in it, including everyone else in the thread. Nothing is read in the background and nothing is read while you type.

On Android, yes — SayThis is on Google Play, as an app with an input method. The iPhone version, an app with a keyboard extension, is coming to the App Store and works the same way.