Swiss keyboard startup Typewise has bagged a $1 million seed round to create out a typo-busting, ‘privacy-safe’ up coming phrase prediction motor created to run solely offline. No cloud connectivity, no info mining threat is the primary concept.
They also intend the tech to work on textual content inputs created on any gadget, be it a smartphone or desktop, a wearable, VR — or a little something weirder that Elon Musk could want to plug into your brain in upcoming.
For now they’ve acquired a smartphone keyboard app that is had around 250,000 downloads — with some 65,000 energetic customers at this position.
The seed funding breaks down into $700K from much more than a dozen local business enterprise angels and $340K by way of the Swiss authorities by way of a mechanism (referred to as “Innosuisse projects“), akin to a study grant, which is paying for the startup to make use of equipment discovering authorities at Zurich’s ETH analysis university to develop out the main AI.
The workforce delicate introduced a smartphone keyboard application late last yr, which consists of some supplemental tweaks (these kinds of as an optional honeycomb format they tout as additional economical and the capacity to edit upcoming phrase predictions so the keyboard promptly groks your slang) to get users to commence feeding in data to build out their AI.
Their principal concentration is on establishing an offline future phrase prediction engine which could be licensed for use wherever end users are texting, not just on a cellular device.
“The objective is to establish a entire world-top text prediction motor that operates entirely on-machine,” claims co-founder David Eberle. “The smartphone keyboard truly is a first use scenario. It is great to exam and create our algorithms in a authentic-lifetime setting with tens of countless numbers of customers. The larger participate in is to convey word/sentence completion to any software that consists of textual content entry, on mobiles or desktop (or in upcoming also wearables/VR/Mind-Computer Interfaces).
“Currently it’s really a lot only Google doing work on this (see Gmail’s automobile completion aspect). Apps these as Microsoft Teams, Slack, Telegram, or even SAP, Oracle, Salesforce would want this kind of efficiency enhance – and at that amount privateness/facts security matters a large amount. Eventually we imagine that each individual “human-machine interface” is, at the very least on the textual content-enter stage, run by Typewise.”
You’d be forgiven for wondering all this seems a bit retro, given the earlier boom in smartphone AI keyboards — these types of as SwiftKey (now owned by Microsoft).
The founders have also pushed unique factors of their present keyboard application — this kind of as the exclusive honeycomb structure — prior to, likely down a crowdfunding route back in 2015, when they have been calling the idea Wrio. But they reckon it is now time to go all in — for this reason relaunching the organization as Typewise and capturing to construct a licensing business for offline up coming phrase prediction.
“We’ll use the resources to create superior text predictions… initial launching it in the keyboard app and then bringing it to the desktop to begin constructing partnerships with related software package distributors,” suggests Eberle, noting they’re doing the job on various enhancements to the keyboard app and also program to devote on marketing to consider to strike 1M lively consumers subsequent calendar year.
“We have a lot more ‘innovative stuff’ [incoming] on the UX facet as very well, e.g. interacting with auto correction (so the user can quickly intervene when it does anything completely wrong — in numerous countries consumers just flip it off on all keyboards because it will get irritating), gamifying the common typing practical experience (huge possibility for young ones/young people, also earning them more knowledgeable of what and how they form), and many others.”
The aggressive landscape close to smartphone keyboard tech, mostly dominated by tech giants, has remaining area for indie performs, is the pondering. Nor is Typewise the only startup considering that way (Fleksy has related ambitions, for a single). Having said that gaining traction vs these types of giants — and about extensive recognized typing strategies — is the tough little bit.
Android maker Google has ploughed useful resource into its Gboard AI keyboard — larding it with characteristics. When, on iOS, Apple’s interface for switching to a third occasion keyboard is infamously disheartening and finicky the opposite of a seamless expertise. Plus the indigenous keyboard gives upcoming phrase prediction baked in — and Apple has lots of privateness credit score. So why would a consumer trouble switching is the problem there.
Competing for smartphone users’ fingers as an indie undoubtedly isn’t quick. Choice keyboard layouts and enter mechanism are always a pretty difficult provide as they disrupt people’s muscle memory and strike cellular end users tricky in their convenience and efficiency zone. Except if the user is affected person and/or stubborn ample to adhere with a frustratingly diverse experience they’ll quickly ditch for the keyboard satan they know. (‘Qwerty’ is an historic typewriter layout turned typing practice we English speakers just just cannot kick.)
Offered all that, Typewise’s retooled concentration on offline following phrase prediction to do white label b2b licensing can make additional feeling — assuming they can pull off the main tech.
And, once more, they’re competing at a information drawback on that entrance vs extra recognized tech big keyboard players, even as they argue which is also a current market possibility.
“Google and Microsoft (thanks to the acquisition of SwiftKey) have a stable know-how in place and have started to offer you textual content predictions exterior of the keyboard a lot of of their competition, having said that, will want to embed a proprietary (hard to develop) or impartial know-how, particularly if their worth proposition is concentrated on privacy/confidentiality,” Eberle argues.
“Would Telegram want to use Google’s textual content predictions? Would SAP want that their clients’ data goes as a result of Microsoft’s prediction algorithms? Which is in which we see our appropriate to get: entire world-class textual content predictions that run on-unit (privacy) and are designed in Switzerland (unbiased setting, no security again doorways, etc).”
Early impressions of Typewise’s up coming phrase prediction smarts (gleaned by via examining out its iOS application) are very low essential (ha!). But it’s v1 of the AI — and Eberle talks bullishly of possessing “world class” developers doing work on it.
“The collaboration with ETH just started out a couple of months back and as a result there are no significant enhancements nonetheless visible in the dwell application,” he tells TechCrunch. “As the collaboration runs until finally the conclude of 2021 (with the chance of extension) the vast the vast majority of innovation is even now to arrive.”
He also tells us Typewise is doing the job with ETH’s Prof. Thomas Hofmann (chair of the Data Analytic Lab, formerly at Google), as perfectly as possessing has two PhDs in NLP/ML and one particular MSc in ML contributing to the effort and hard work.
“We get distinctive rights to the [ETH] technology they really do not keep fairness but they get paid by the Swiss govt on our behalf,” Eberle also notes.
Typewise says its smartphone app supports additional than 35 languages. But its next word prediction AI can only take care of English, German, French, Italian and Spanish at this place. The startup claims far more are being added.