Give someone back
their voice.
Kural uses eye tracking and AI to let ALS patients speak in their own cloned voice — with just 3 eye movements per sentence.
ALS takes the voice.
We give it back.
Today's AAC devices were designed for motor control, not for someone whose eyes are their only window left.
"The eyes are the last thing ALS takes.
We built for that moment."
Three steps. One voice.
Bank your voice
Record sentences while you still can. Our pipeline trains a voice model on AWS SageMaker that captures your exact timbre, cadence, and inflection — not a generic synthetic voice.
Set up your profile
Tell us about your communication style, your family, your humor. Kural learns who you are — so the sentences it suggests sound like you, not a machine.
Speak with your eyes
Gaze at intent tiles. Kural generates full, contextually aware sentences via Amazon Bedrock. Kural speaks the sentence back in your cloned voice.
Powered by AWS
Kural runs on a fully serverless AWS stack — eye input on iPhone feeds directly into API Gateway, Lambda, and Bedrock with no edge hardware required.
Architecture overview
You set it up.
It adapts over time.
Kural is designed so caregivers can get started without technical support — and so the device evolves alongside the patient's changing needs.
- Self-service onboarding — record, upload, done in under an hour.
- Caregiver dashboard to update patient preferences, topics, and family names anytime.
35,000 people in the US live with ALS. Most will lose their voice. Almost none have a tool that preserves it.
We're a team of University of Washington computer science students who believe that the technology to solve this problem already exists — it just hasn't been applied where it's needed most. Kural is our attempt to close that gap. We're building for the people who deserve better tools, not for the people who make headlines.