It would be logical to implement autocomplete in messengers based on message analysis – in your chats, your friends’ chats, and people with similar profiles (same language, interests, etc.), and to train swipe keyboards, autocorrect, and auto-replies to use this analysis data in making decisions on which word or phrase to insert in autocomplete, autocorrect, or auto-reply. It seems not too complicated, yet somehow no one is doing it. Only Google in their email and LinkedIn in their quasi-messenger are trying with autocomplete. With such training, voice recognition for entering text messages could be made much better than it is now. It would understand slang and a mix of English and Russian words. Translators could do better, as they could be trained on users’ messages. We are generating megabytes of data by typing to each other.
Unfortunately, all the major players in this market consider users’ messages only as raw information for tailoring relevant ads. The messengers themselves don’t seem to change at all.
No messenger supports plugins – the way it’s done in Atom or IntelliJ Idea, or even as in Chrome. It should have a catalogue where you choose, click install, and some useful function gets enabled. But plugins don’t have access to the entire message database, whereas messenger owners do – and they do nothing.
Hi, I'm Rauf Aliev. 