WebAble
The WebAble ecosystem

Three surfaces. Not a catalogue.

This page used to list six products with six signup forms. Most of them were one application. Able is a single macOS app with several ways in, the browser extension runs where a desktop app cannot reach, and the widget puts the extension’s panel on a site for its visitors. That is the whole list.

Inside Able

The modes are not downloads.

GlassLayer, EchoControl and AutoFlow were never separate applications. They are ways into the one app, reached by a chord or by asking, and they arrive with it.

  • Guide

    Always docked

    Ask about the window in front of you. It reads the controls that window publishes, talks you through the next step, and moves the pointer onto the control it names.

  • Ask in private

    Cmd+Shift+G

    A surface excluded from screen recording and screenshots, which can also transcribe the system audio of a meeting while it runs.

  • Dictate

    Hold Cmd+Shift+D

    Speak into whatever application holds the caret. A personal dictionary fixes the words your recogniser gets wrong, and the text never enters clipboard history.

  • Skills

    Say the name

    Save an instruction under a name, then call it by saying that name. It ships with none, and nothing is recorded or replayed.

A long term memory is the one thing that used to be advertised here and does not exist. Able keeps a rolling ten minute record of the conversation and then forgets it, on purpose.

What holds it together

The same idea in three places.

Adapt the interface to the person in front of it, rather than waiting for the interface to be rebuilt. What the surfaces genuinely share is below. Where they do not share anything, we say so.

It reads controls, not pixels

Able asks the window what it is showing: the name, role and exact rectangle of each control, the same list a screen reader uses. That is why the pointer lands on the control rather than near it, and why a screenshot is only the fallback.

The panel is one codebase

The browser extension and the website widget render the same panel from one shared package, so a tool fixed in one is fixed in the other. Able is native macOS and does not share that code.

Speech stays on the machine

Recognition and the voice that reads answers back both run locally in Able, for dictation and meeting transcription alike. No audio is uploaded. Every window Able owns is hidden from screen recording.

How we're rolling this out

One surface at a time.

The browser extension is the front door. It is in public beta, it is free, and getting it widely used is what earns the right to put anything else in front of a bigger audience.

Able is the newer surface and the more invasive one: it asks for Accessibility, Screen Recording and the Microphone, so it goes out by email, to people who want to run a pre-release build, rather than through a download button. There is one list for the whole app, because there is one app.

What is not finished stays labelled as not finished. Every gap we know about is on the product page, in the same words the repository uses.

Start where you work.

Able for the desktop, the extension for the browser. One of each, not six of anything.