Say hi. A human writes back.

Any camera that saves video files we can read: Reolink, Wyze, Ring, Nest, Arlo, Eufy, or a plain IP camera writing to a NAS. MP4, MOV, and MKV are all fine. We run it on a wall-mounted Reolink ourselves. We're also building a small Raspberry Pi camera kit for homes that don't have a camera yet.

Face recognition and identity matching run on hardware in your home. Your child's face embeddings never leave it.

Clips that match your child are sent over an encrypted connection to a vision model that writes the observation, and the observation is what we store. We're working to run that step locally too. Clips you haven't bookmarked expire on a fixed schedule.

Three to five photos of your child's face from different angles, in decent light. The system builds several reference centroids from them, and gets better as it sees more of your footage.

No, and that's deliberate. PeetaBoo writes factual observations: "attempted new words today", "climbed onto the armchair for the first time". It never suggests a medical condition or concern. The notes are the kind of thing you'd share with your pediatrician at a routine visit, and that's who they're written for.

Yes. Each child gets their own profile, enrolment, and journal. Siblings and parents are enrolled as known faces so they're never mistaken for one another.

Only people you invite to your household. Admins run the household; viewers, like grandparents, see everything or nothing. Invitations are single-use and expire after 48 hours. Removing someone cuts their access immediately.

We're in private beta with a small group of families. Join the waitlist and we'll be in touch as we open up more homes. Early families help shape the product.