Skip to main content

Open Independence · Open Builder

You do not need to learn to code before you are allowed to build.

Open Builder is a practical program for turning ideas into working apps, games, websites and tools using AI development agents. It is designed first for blind and visually impaired builders, and it works for anyone with little or no coding experience.

This is a building lab, not a video course

Every lesson asks you to do something: describe, build, test, break, fix, decide. You will learn to direct an AI agent, understand what it changed, verify the result without needing to look at a preview, and improve your instructions when the result is wrong.

  • Direct an AI agent

    Write instructions precise enough that an agent cannot guess wrong.

  • Understand the change

    Read what was built well enough to supervise it.

  • Test what happened

    Verify behaviour with the keyboard and with structure, not with a screenshot.

  • Fix what breaks

    Diagnose, describe, roll back, and stop failed-fix spirals.

  • Build accessibly

    Accessibility is taught as part of building, not as a clean-up pass.

  • Ship something real

    Finish with a public project another person can actually use.

What the course centres on

The work is sequenced so each part earns the next. These are the parts that matter most.

  • Talking to an AI coding agent

    The core skill: instructions clear enough that the result is what you asked for, and revision when it is not.

  • When things go wrong

    A repeatable recovery routine, so a broken build is a problem you work through rather than a dead end.

  • Build accessibility in

    Keyboard operation, structure, focus and announcements checked as you build, using tools you can operate without sight.

  • Your own project, published

    The course ends with something real: your project, shipped, tested live, and usable by a stranger.

Accessibility is the foundation

The platform targets WCAG 2.2 AA throughout: full keyboard operation, visible focus, semantic structure, descriptive errors, announced status changes, text alternatives, adjustable text size, high contrast, reduced motion, and no auto-playing media. No lesson requires vision to complete, and no exercise asks you to look at a preview to find out whether your project works.