KIN Access

KIN Access is AI-powered switch access for people with severe motor disabilities like ALS, spinal cord injuries, and cerebral palsy. Many of these users control a computer with a single switch, pressed with a cheek, head tilt, or puff of air. The standard method, switch scanning, highlights every item on screen one at a time, so replying to one email takes at least 23 presses. KIN Access uses Meta’s Muse Spark to predict the 4 actions the user most likely wants next. The user picks one with a single press. Risky actions like send and delete always require a confirmation step, with Cancel highlighted first. If none of the 4 fit, “Something else” falls back to full scanning, so the user is never stuck. In an automated evaluation that plays a perfect switch user with fresh Muse Spark calls, KIN Access used 63% fewer presses than the best possible traditional scanning and beat it in 24 of 24 runs. In Vision mode, where Muse Spark sees only a screenshot with no text description, it still used 59% fewer presses and beat traditional scanning in 24 of 24 runs. The AI only suggests. Ordinary code validates every suggestion against what is actually on screen before anything runs. KIN Access is an accessibility tool for computer control and gives no medical advice. All code was built during the hackathon.