Loops Houselullwatch

NEAR HACKATHON
Links
Team
1 member- TOOwner
Tolulope
Overview
The problem: in a first pregnancy, the hardest call is "is this it?" In a study of 1,656 first-time mothers, 1 in 3 went to hospital before contractions were 5 minutes apart, and those early arrivals had more caesareans (24% vs 16%). Leave it too late and the journey becomes the risk. And the apps that could help collect the most private data there is: Mozilla flagged 18 of 25 reproductive health apps, and in 2025 a US jury found Meta liable over Flo app pregnancy data.
What I built: Lullwatch, a private labour companion. She times contractions by tap or voice. Fixed, tested rules (5-1-1 by default, her midwife's numbers if different) forecast when she'll meet the rule, subtract her journey, and say "start packing" then "leave by" a time. Any warning sign overrides with "Go now". Tapping Call midwife opens a script for the phone call, written by the AI and checked against the timer. It installs and works offline, and contractions never leave her phone.
How NEAR fits: the AI runs on a TEE-hosted model on NEAR AI Cloud. Every message is end-to-end encrypted to the attested model key. Her browser verifies the hardware itself with Intel (TDX quote) and NVIDIA (NRAS) using a fresh nonce, and checks a signature on every reply. The server keeps nothing.