Android keyboard ditches keys entirely, predicts what you mean

The Register / 3/31/2026

📰 NewsSignals & Early TrendsTools & Practical Usage

Key Points

  • The article describes an Android touch keyboard approach that removes traditional key-by-key input and instead predicts what the user is trying to type.
  • It highlights that the design is aimed especially at blind tablet users, while also appealing to sighted users who want faster, easier text entry.
  • The piece frames the keyboard as a shift toward predictive, intention-based typing rather than manual key selection.
  • It reports on the product/feature’s emergence and growing adoption rather than a standalone technical tutorial or research explainer.

Android keyboard ditches keys entirely, predicts what you mean

Aimed at blind tablet users, although it's winning sighted fans too

Tue 31 Mar 2026 // 09:27 UTC

TapType is a new Android keyboard that's invisible. You can't see it – but that's OK, neither can its developer nor some of its target users.

Developer Aaron Hewitt released TapType 2.0 earlier this week. It's a keyboard for Android devices that tracks the relative positions of where your fingertips tap the screen and extrapolates which keys in the QWERTY layout you were aiming for.

The Reg FOSS desk was interested in this for a few reasons. First – and it's a big one for this vulture personally – we're not very happy with the current state of Android keyboards, as we have described at some length before. We installed TapType on our backup phone, but to be honest, we didn't get very far. Once we tapped a text field, the bottom part of the screen simply went blank.

There's an invisible QWERTY keyboard in there, and tapping around on it occasionally generated letters – but we can't touch-type on our phone screen, because it's well under half the width of the four fingers of one of our hands, excluding the thumb.

But we were also intrigued by the rationale for the design, and we greatly enjoyed Hewitt's lengthy and impassioned blog post about developing it, "I Made a Keyboard Nobody Asked For: My Experience Making TapType."

He has legitimate reasons for his strong feelings about using touchscreen devices, and they're the same reasons that TapType doesn't display anything on the screen. Hewitt is blind. Not being able to see what's on the screen of a device that is all screen does make it a lot harder.

Apple tends to take accessibility a lot more seriously than most folks, particularly in the Linux world, and that's also something we wrote about for Global Accessibility Day 2025, comparing Apple's efforts with that of other vendors. Most of our blind friends and acquaintances tend to use Apple devices for that very reason.

Hewitt doesn't – he favors Android, and he uses Linux as well. Last year, he wrote an excellent four-part blog post about the experience of trying to use desktop Linux as a blind person, under the collective title of "I Want to Love Linux. It Doesn't Love Me Back." We recommend reading all the installments. The first three are very damning, and they set the context for our Apple story.

  1. Built for Control, But Not for People
  2. The Audio Stack Is a Crime Scene
  3. Interlude – A Thank You, Where It's Due

Then the fourth came along a whole month later, with a surprise ending: "Wayland Is Growing Up. And Now We Don't Have a Choice."

So when the same author released an Android keyboard a year later, we were very curious. We asked him why he'd picked Android instead of iOS.

"Most of the time, I don't need a phone to be a phone," he told us. "I do a lot of my work from my phone. TapType was partly developed on the phone, on the couch. I need my phone to be as multi-purpose as it can possibly be, and iOS doesn't give me that."

We've found that too. The FOSS desk uses macOS on the desktop, where we can choose our own keyboard and mouse and FOSS apps. In comparison, iOS is far more locked down, and its apps are sandboxed. You can't simply save a file from one app and open it in another – the OS is specifically designed to prevent that kind of thing.

Hewitt continued: "I also like new toys. I want things like foldable form factors, and Apple doesn't do them – but Android vendors do."

Even so, he's not a fan of the Android accessibility or development tools. "Oh, TalkBack is crap. But the thing is, on Android, I can write my own, or get one from a third party – and either way, it's way more customizable than VoiceOver."

He had harsher words for the Android development kit, and told us that it's so lacking in accessibility that he has to use real hardware for testing. He has stacks of old Android phones around his house, and flashes different versions of Android onto them for testing. An old Samsung Galaxy S7 is particularly helpful for this, as it can run so many different versions of Android.

One surprise was that Hewitt isn't as much of an AI skeptic as we had expected – especially given his February post, "The Slow Death of the Power User."

He told us that he finds AI tools very useful, especially for tasks like describing what something looks like on screen. "I like AI. I use it for search. It can be very handy to be able to ask 'what is the latest Compose bomb, and what can I do about it?'" he said, referring to Jetpack Compose, the Android UI development tool. "Oh, it's awful, it's so bad. I found a drop-down labeled as an edit text control!"

Part of the motivation behind TapType was the rise and fall of Fleksy. This iOS keyboard launched in 2012 as an accessible keyboard for blind users, and also a very fast one for sighted users. In 2014, The Register reported on its world speed record. But, as Hewitt's description of the development of TapType describes, Fleksy found that the larger market for sighted users was more lucrative, and the accessible version was spun off as a separate product, Fleksy VO, based on an older version of the code, which didn't get as much attention as the version for the sighted market. Some users were not impressed, as Jonathan Mosen described in "I Uninstalled Fleksy Today, and Want to Tell you Why."

So he plans for TapType to remain mainly aimed at a user base that doesn't need it to have a visible UI. However, he has been very surprised by its uptake.

"So far, it's more popular with sighted users than blind ones! But saying that, the uptake with blind users has been very, very good. It's exploded. I expected one or two people would find it, and then that I'd never hear from anyone ever again. Saying that, it's for me, and I didn't make it for anyone else."

Hewitt did warn us that if you attempt to install the APK file from his GitHub, you'll need to enable additional software sources, and even then, "expect big scary warnings from Play Protect. It is going to tell you 'This will steal your data!' and so on. It won't. I don't want anyone's data. I am not interested. But it's funny because I have done much more unsafe stuff in the past, Android apps that let you send any raw HTTP queries you want and more, and Play Protect doesn't flag that up at all."

It reminds us of Canonical's problems with rogue Snap Store packages. Automated software checks can't usefully tell malicious apps from safe ones, and it's not getting any better.

Colin Hughes, a past Reg contributor and disability rights advocate who runs news outfit Aestumanda.com, also gave it a go. He remarked that: "As a sighted person, I found a completely blank keyboard surprisingly disorientating. It made me realize just how much I rely on visual cues and established muscle memory from a conventional keyboard.

He added that although he struggled to "orient myself on what is, visually, a blank surface," he could "immediately see the logic of it and why it could be so useful to the people it is really designed for."

"One useful reminder from trying it is that accessibility is not one thing. I found the swipe gestures quite natural, but because of my physical disability I could not perform the two-finger command gestures. A tool can be transformative for one group of disabled people while still creating barriers for another."

Hughes notes that what really stayed with him was "the story behind it." "TapTypematters not just because of what it does, but because of why it had to be made at all. It exists because one developer got frustrated enough to solve a problem the mainstream industry had clearly failed to treat as a priority. There are obvious parallels there with voice access, which I rely on, and the disappointing state of built-in dictation on phones and laptops. Too often, disabled people are left making do with inferior tools, or with tools that were never truly built for them, until an individual developer steps in and delivers what much larger companies could have done years earlier.

"The contrast is hard to ignore when three- or four-person teams such as those behind Aqua Voice and WhisperTyping are producing more ambitious voice dictation experiences than companies on the scale of Apple and Microsoft.That is why I think TapType is more significant than just being an interesting keyboard. Its existence is evidence of a wider market failure. When a specialist accessibility tool appears and people respond to it quickly, that is not only a compliment to the developer. It is also a sign that the wider industry has left an important need unmet forfar too long."

TapType is an interesting experiment. There's significant prior art in this area. A friend of ours used to use a remarkable Twitter client called Qwitter, which is similarly invisible: it doesn't have a window, and shows nothing at all on screen – its UI is typing and speech.

There's also been considerable research into trying to predict what someone is typing simply from the relative finger positions on a flat surface. FingerWorks did some pioneering work in this area before Apple bought the outfit and its products disappeared. Its fabled trackpad that was also a keyboard never appeared as a product in its own right. Another recent research effort in this direction was also called TapType, which we hope doesn't cause Mr Hewitt any problems. ®

More about

TIP US OFF

Send us news