Team 4

Milo ai

Milo is the personal AI agent built for people with disabilities: if you can move your head, sign or speak, you can use it.

Check your own project

Community votes recognize the build and do not affect official judging.

Review the project

Start with the source code, then open the demo or video if available.

View GitHub repository
Project description
Milo is a personal AI agent for people who can't reliably type, click or speak: people with limited hand or arm movement, people for whom speech is slow or tiring, and Deaf users. Most agents start with a text box. Milo starts with whatever the person can do, across four inputs: text, voice, head and sign. In head mode, the webcam turns your nose into a pointer. You pick only the first letter of each word, Milo predicts the sentence from memory of what you usually say, and a nod confirms. "Can you help me?" takes 3 moves instead of 12 letters. Sign mode reads fingerspelling plus YES, NO, HELP and STOP. Camera frames never leave the device. Then it acts. The demo task is one wheelchair users face: getting the right chair bought and paid for. • Fit check: every chair is tested against her body and doorways, with the numbers shown. 2 of 8 fit, and the insurer's default chair does not. • Insurance: reads the denial and the policy, cites each clause, and drafts the appeal. A simulated strict reviewer attacks the draft, Milo revises it, and it is filed only after she confirms. • Checkout: refuses chairs that won't fit and sends anything over $500 to her caregiver for approval. • Speaks for her: chosen phrases are said aloud in a human voice. Every action lands in Receipts: the input mode, what Milo understood, confidence, what she confirmed, evidence with sources, and whether each service call was live. Demo data is fictional, and filing and checkout are sandboxed.
Tools used