Neverclick is the kind of Windows utility you do not ignore once you try it. Developer Lazo Velko built a free, offline app that lets you perform mouse actions on practically any object on your screen using only keyboard shortcuts. No signup. No ads. No data collection. And it is not even trying to play nice with the traditional “accessibility APIs” approach. Instead, it uses computer vision to analyze pixels, identify individual elements, and then click them for you.

The practical consequence is immediate: after hitting Ctrl + Enter, or Alt + Enter for Ctrl + Left Click, your screen gets covered in two-letter keyboard “hints” (ek, af, jj, and more). Type a hint and the cursor clicks in that exact spot, whether it is a button, a menu option, or even the middle of a word. You can use numbers to go from a word hint to a particular letter, switch windows via keyboard shortcuts, and choose whether hints appear only in the active window or across the entire screen. Drag and drop and highlight are not supported yet, but the core promise, keyboard-only targeting of on-screen elements, works now.

So why does this matter beyond accessibility geeks? Because Neverclick is built around a specific design decision that directly challenges how Windows software usually thinks about “who can control what.” Velko says he did not build it with accessibility APIs at all. He also says he has “had a poor experience with accessibility apis,” calling them “clunky, slow, and unpredictable.” In that same Hacker News discussion, he argued that his pixel-based approach avoids that unpredictability, saying he can have users on 10-year-old hardware where the computer vision “runs instantly,” while Windows UI Automation is “super laggy.” He claims his system is built without third-party libraries and is small, with the entire app around 40 MB of disk space, and using around 200 MB of memory on a typical 1080p monitor when in use. He also notes it does not run continuously, so it should not eat CPU unless summoned by the key combo.

Underneath the technical framing is a user-experience problem that many executives have seen in enterprise IT and product accessibility alike: the tools that are supposed to help are often the ones that get stuck. Accessibility APIs are meant to be a bridge between the UI and assistive input, but if they are inconsistent across software, or too heavy for certain machines, the “bridge” becomes a bottleneck. Velko’s alternative does not depend on the underlying app exposing structured accessibility information. It just looks at what is on the screen, then guesses where your target is. That is the whole tradeoff. When it works, it is remarkably universal. When it does not, it does not fail gracefully in the way an accessibility tree-based approach might.

That also explains why Velko explicitly acknowledges the limits. The Register notes that Neverclick will not work with software that was not built with accessibility in mind, because even though he avoided accessibility APIs, software behavior still affects what is realistically clickable or identifiable. In other words, pixel vision helps avoid one class of dependency, but it does not magically solve every UI that refuses to behave like a stable set of elements. For product and platform leaders, this is a useful reminder that “assistive control” is not one feature. It is a chain, and any weak link can break the user journey.

Then there is the messy reality of shortcuts and environments. Neverclick uses Ctrl + Enter as a default trigger for the overlay, and The Register reporter Avram Piltch saw a conflict: Ctrl + Enter is supposed to send an email in Gmail, but with Neverclick running, it brings up the overlay instead. The hints overlay itself can be overwhelming too, because it covers menus and icons while it is active, and the documentation suggests you have to click to make it go away. These are not deal-breakers for a personal productivity tool, but they are exactly the kinds of frictions that matter if you are thinking about rolling similar functionality into broader workflows or enterprise environments.

Neverclick is Windows-only for now. Velko told The Register that a macOS version is “in the works,” but he does not have a timeline, because he is “overworked” and still needs to release a few features for Windows. Those include window dragging and text highlighting by selecting two different hints, then triggering whatever text editor is being used to highlight what is between them. He also told the publication he has not published the full source code yet; the GitHub repo is currently for issue reporting only.

From a strategic angle, Neverclick is less about whether one keyboard utility is cool, and more about what approach wins when you need “mouse-like control” without asking every app developer to comply. If Velko’s claims hold up in the wild, pixel-based control could pressure platform teams to think harder about the performance and predictability of accessibility layers like Windows UI Automation. At the same time, product teams should notice the second-order implication: even when an assistive system is technically universal, it can still collide with app shortcut conventions and overwhelm users with opaque overlays. The executives watching interface reliability should take the signal seriously, because the next wave of control software will not just be about accessibility. It will be about robustness under real-world constraints: old hardware, messy UIs, shortcut collisions, and the basic question of whether “helpful” input methods actually behave like help.