What you need to run it

This is an R&D project, not a product you can buy yet. If you want to follow along hands-on, here’s an honest picture of what it takes today — and where things are still being figured out.

The Hub: The rev-1 target is a generic Linux box — an x86 mini-PC or an ARM single-board computer — running Linux and Docker, not a Mac. Onboard AI runs as a separate Jetson sidecar that publishes what it sees back to the hub over the LAN; the hub itself is never migrated to it. Mac and iPad are LAN clients, not the hub — they update via the App Store, and the hub only displays to them. For now, dev and demo run on a MacBook, inside a Linux/arm64 Docker container. (Exact minimum RAM/storage/CPU isn’t pinned down in the architecture docs — confirm before you shop.)
Apple ecosystem: The apps ship as React Native, for web and iOS, talking to the Rust core over wasm (web) or UniFFI (iOS) — so a Mac or an iPad is the native path once apps run as separate clients on the boat’s network. Honest caveat: that separate-client path — a second device on the boat’s WiFi reaching the hub over the LAN — is a decided architecture, not yet built. Today, hands-on means running the app in-process alongside the hub, not yet a phone or iPad talking to it over the water.
NMEA gateway: The code already decodes NMEA 2000 and NMEA 0183 sentences into the canonical model — that part’s real and tested. What isn’t built yet is the live wiring to a physical boat: reading actual bytes off a USB or UDP gateway is still in progress. The planned/target gateways are an NGT-1 or a Yacht Devices YDNU-02 (both USB-serial) or a YDWG-02 (over UDP) — named because they’re what the ingest layer is built toward, not because one works plugged in today. If you don’t have a boat wired up, you don’t need one to follow along: there’s a built-in NMEA/AIS simulator that feeds the same decoders with no hardware at all.

Back to The work Get updates