Hearth is a local-first AI hub that watches, listens and understands the nursery — so parents can rest, and their child's data stays at home. No cloud, no subscription required, no video that ever leaves the house.
Give every parent a calm, trustworthy second set of eyes on their baby — one that runs in their own home, answers to them alone, and never turns their child into a data product.
New parents are sold anxiety. The baby-monitor market has moved from a simple radio to a camera, an app, a subscription, and a cloud that stores months of nursery video and biometric data on a vendor's servers. We believe the intelligence belongs in the home — a hub the size of a paperback, watching the crib, listening for cries, tracking the room, and working with the vital-sign wearables families already own, with every frame of video and every heartbeat processed and stored inside the house.
The technically capable minority already solve this with weekends of open-source home-automation tinkering. Everyone else has no alternative.
Leading connected monitors stream nursery video to remote servers by default. Core features stop working during outages, subscriptions rise, and some products have been bricked when the vendor moved on.
Camera, pulse-ox sock, room thermometer and sleep log each have their own app, none of which talk to each other. A parent gets four notifications about one event, or none about the one that mattered.
Single-sensor alarms fire on a loose sock or a passing car. Parents learn to ignore them, which defeats the purpose of having a monitor.
Infant biometric and video data is among the most sensitive a household produces, yet it is routinely retained, analysed, and in some cases shared under broad consent language.
Hearth connects to a nursery camera, a room sensor, and optionally an Owlet Dream Sock, and runs all analysis on the device.
Knows whether the baby is asleep, restless, awake, rolled, or climbing — from the camera, on-device.
Distinguishes a brief fuss from a sustained cry, a cough from background noise, and tells you which.
Tracks temperature, humidity, light and noise; can drive a fan or air conditioner to keep the room in the safe range.
Flags face-covered, uncovered, and — under controlled lighting — noticeable skin-tone change versus the child's own baseline, always with a snapshot for the parent to judge.
Pulls heart rate, oxygen and skin temperature from the family's existing sock and cross-checks them against movement and sound, so a loose sock reads as a loose sock.
Fuses every signal into a single state and notifies on meaningful changes, not raw readings. Adult-in-room and quiet-hours awareness built in.
"How did she sleep last night?" answered by a small language model running on the hub, from the day's event log.
Video never leaves the LAN. Privacy is a property of the design, not a promise in a terms-of-service document.
Works with any standard IP camera and any common room sensor. Parents aren't locked into our hardware, and we aren't locked into any one supplier.
A sub-$200 mini PC or single-board computer can now run real-time object detection, pose estimation and audio classification simultaneously — three years ago this required a cloud GPU.
Camera-vendor breaches and subscription backlash have made "local" a feature consumers ask for by name, well beyond the enthusiast niche.
Home Assistant, Frigate and the surrounding ecosystem have proven the architecture with millions of installs. Our job is to productise it for a specific, high-value moment in a family's life, not to invent it.
Owlet is in hundreds of thousands of nurseries. No product today combines its vitals with camera context. Community integrations show the demand; a supported product does not yet exist.
Our first customer already owns, or is about to buy, a premium connected monitor and a pulse-ox wearable — concentrated, reachable through the same channels that sell the wearables, and highly motivated by privacy and reliability. The model is intentionally the inverse of the incumbents: this is cheaper to support, aligns with the privacy promise, and earns trust that compounds into referrals in a category driven almost entirely by parent-to-parent recommendation.
A pre-configured hub with accelerator, sold at a healthy margin over commodity components. One-time purchase; the customer owns it outright.
Hub plus a vetted camera and room sensor for parents who want one box.
Not required for core function. Covers secure remote access, cloud-free backup coordination between family devices, priority support and early model updates.
Co-marketing and, where possible, supported integration agreements with wearable and camera vendors.
Expansion path: toddler and child-room monitoring on the same hub, elder care (the same fusion of movement, sound, environment and wearable vitals maps directly onto ageing-in-place monitoring), and pet or general home awareness as a retention play once the hub is in the house.
Working hub on reference hardware: movement, cry, environment, Owlet cross-check, single notification stream. Founding-team nurseries.
Complete / in progress50–100 households from parent communities and wearable-owner groups. Measure false-alert rate, setup time, and retention.
Next 6 monthsPre-configured hub and starter bundle; companion app on iOS and Android; direct-to-consumer and one retail partner.
12 monthsSupported integration with at least one wearable vendor; camera vendor bundle.
12–18 monthsToddler profiles; elder-care pilot.
18–24 monthsThis site is the pitch. The full white paper covers the market, business model, technology layers, and roadmap in detail.
Hearth moves from working prototype to pilot and launch over the next year: hardware design and certification for the hub, the companion app, a pilot programme with real families, and the first partnership conversations with wearable and camera vendors.