Published On: Fri, Jul 24th, 2020

Typewise taps $1M to build an offline subsequent word prophecy engine

Swiss keyboard startup Typewise has bagged a $1 million seed spin to build out a typo-busting, ‘privacy-safe’ subsequent word prophecy engine designed to run wholly offline. No cloud connectivity, no information mining risk is a simple idea.

They also intend a tech to work on content inputs done on any device, be it a smartphone or desktop, a wearable, VR — or something weirder that Elon Musk competence wish to block into your mind in future.

For now they’ve got a smartphone keyboard app that’s had around 250,000 downloads — with some 65,000 active users during this point.

The seed appropriation breaks down into $700K from some-more than a dozen internal business angels; and $340K around a Swiss supervision by a apparatus (called “Innosuisse projects“), same to a investigate grant, that is profitable for a startup to occupy appurtenance training experts during Zurich’s ETH investigate university to build out a core AI.

The group soothing launched a smartphone keyboard app late final year, that includes some additional tweaks (such as an discretionary honeycomb blueprint they surveillance as some-more efficient; and a ability to revise subsequent word predictions so a keyboard fast groks your slang) to get users to start feeding in information to build out their AI.

Their categorical concentration is on building an offline subsequent word prophecy engine that could be protected for use anywhere users are texting, not usually on a mobile device.

“The idea is to rise a world-leading content prophecy engine that runs totally on-device,” says co-founder David Eberle. “The smartphone keyboard unequivocally is a initial use case. It’s good to exam and rise a algorithms in a real-life sourroundings with tens of thousands of users. The incomparable play is to move word/sentence execution to any concentration that involves content entry, on mobiles or desktop (or in destiny also wearables/VR/Brain-Computer Interfaces).

“Currently it’s flattering most usually Google operative on this (see Gmail’s automobile execution feature). Applications such as Microsoft Teams, Slack, Telegram, or even SAP, Oracle, Salesforce would wish such capability boost – and during that spin privacy/data confidence matters a lot. Ultimately we prognosticate that each “human-machine interface” is, during slightest on a text-input level, powered by Typewise.”

You’d be forgiven for meditative all this sounds a bit retro, given a progressing bang in smartphone AI keyboards — such as SwiftKey (now owned by Microsoft).

The founders have also pushed specific elements of their stream keyboard app — such as a particular honeycomb blueprint — before, going down a crowdfunding track behind in 2015, when they were job a judgment Wrio. But they reckon it’s now time to go all in — hence relaunching a business as Typewise and sharpened to build a chartering business for offline subsequent word prediction.

“We’ll use a supports to rise modernized content predictions… initial rising it in a keyboard app and afterwards bringing it to a desktop to start building partnerships with applicable program vendors,” says Eberle, observant they’re operative on several enhancements to a keyboard app and also devise to spend on selling to try to strike 1M active users subsequent year.

“We have some-more ‘innovative stuff’ [incoming] on a UX side as well, e.g. interacting with automobile improvement (so a user can simply meddle when it does something wrong — in many countries users usually spin it off on all keyboards since it gets annoying), gamifying a ubiquitous typing knowledge (big event for kids/teenagers, also creation them some-more wakeful of what and how they type), etc.”

The rival landscape around smartphone keyboard tech, mostly dominated by tech giants, has left room for indie plays, is a thinking. Nor is Typewise a usually startup meditative that approach (Fleksy has identical ambitions, for one). However gaining traction vs such giants — and over prolonged determined typing methods — is a wily bit.

Android builder Google has ploughed apparatus into a Gboard AI keyboard — larding it with features. While, on iOS, Apple’s interface for switching to a third celebration keyboard is infamously frustrating and finicky; a conflicting of a seamless experience. Plus a local keyboard offers subsequent word prophecy baked in — and Apple has copiousness of remoteness credit. So because would a user worry switching is a problem there.

Competing for smartphone users’ fingers as an indie positively isn’t easy. Alternative keyboard layouts and submit apparatus are always a really tough sell as they interrupt people’s flesh memory and strike mobile users tough in their comfort and capability zone. Unless a user is studious and/or realistic adequate to hang with a frustratingly opposite knowledge they’ll shortly embankment for a keyboard demon they know.  (‘Qwerty’ is an ancient typewriter blueprint incited typing robe we English speakers usually can’t kick.)

Given all that, Typewise’s retooled concentration on offline subsequent word prophecy to do white tag b2b chartering creates some-more clarity — presumption they can lift off a core tech.

And, again, they’re competing during a information waste on that front vs some-more determined tech hulk keyboard players, even as they disagree that’s also a marketplace opportunity.

“Google and Microsoft (thanks to a merger of SwiftKey) have a plain record in place and have started to offer content predictions outward of a keyboard; many of their competitors, however, will wish to hide a disdainful (difficult to build) or eccentric technology, generally if their value tender is focused on privacy/confidentiality,” Eberle argues.

“Would Telegram wish to use Google’s content predictions? Would SAP wish that their clients’ information goes by Microsoft’s prophecy algorithms? That’s where we see a right to win: world-class content predictions that run on-device (privacy) and are done in Switzerland (independent environment, no confidence behind doors, etc).”

Early impressions of Typewise’s subsequent word prophecy smarts (gleaned by around checking out a iOS app) are flattering low pivotal (ha!). But it’s v1 of a AI — and Eberle talks bullishly of carrying “world class” developers operative on it.

“The partnership with ETH usually started a few weeks ago and so there are no poignant improvements nonetheless manifest in a live app,” he tells TechCrunch. “As a partnership runs until a finish of 2021 (with a event of extension) a immeasurable infancy of creation is still to come.”

He also tells us Typewise is operative with ETH’s Prof. Thomas Hofmann (chair of a Data Analytic Lab, before during Google), as good as carrying has dual PhDs in NLP/ML and one MSc in ML contributing to a effort.

“We get disdainful rights to a [ETH] technology; they don’t reason equity though they get paid by a Swiss supervision on a behalf,” Eberle also notes. 

Typewise says a smartphone app supports some-more than 35 languages. But a subsequent word prophecy AI can usually hoop English, German, French, Italian and Spanish during this point. The startup says some-more are being added.

About the Author