I’ve run Home Assistant for years, but reliable family presence has always been the weak link. Companion app works for techies, but not for real families. Router-based methods broke long ago. I wanted something that “just works” without constant maintenance.
Unlike router/MAC-based presence, this doesn’t break when your phone randomizes its MAC address — presence comes from the app reporting in, not network detection.
Family members install the PositionGuard app and join your group. That’s it. The integration creates clean native device_tracker + binary_sensor entities. Standard Home / Away + named areas. Works across multiple HA instances too — each instance runs independently, same or separate API keys. Architecture’s proven on multi-hub Hubitat setups; haven’t personally verified it across multiple HA instances yet, so treat that as confident-but-unconfirmed for now.
Note: Uses the hosted PositionGuard service (free for typical families). The integration depends on the hosted service, not local-only — presence syncs through the cloud, and HA only ever sees which named area someone’s in, never exact coordinates. If local-only is a dealbreaker, that’s fair, and worth knowing before you install.
Feedback welcome — including from anyone running multiple HA instances.
That's a really good idea — and you've put your finger on exactly the right friction point. Long-time HA users have years of Zones built up, and asking people to re-create that geography by hand in another app is real transition cost.
Honest current state: the integration is read-only today (by design — group and area management lives in the app), so there's no write path for this yet. But an import of HA Zones as PositionGuard areas is a genuinely good fit for the roadmap, and it wouldn't cross the privacy boundary since defining areas is an owner action, not a member-sharing one.
One question back so I build the right thing: would a one-time import (pull my Zones in when I set up) cover it for you, or do you actually want ongoing sync — Zones changed in HA staying matched in PositionGuard? The first is much simpler and I suspect it's what most people need for the transition you're describing, but tell me if I'm wrong.
Quick update: added multi-instance / multi-site notes to the README — running the integration across separate HA instances (e.g. main home + vacation property) is architecturally supported, and it’s been tested successfully in a couple of multi-hub Hubitat setups already.
Honest caveat: I haven’t personally run it across multiple HA instances yet, so I’m not calling it fully verified on HA specifically — but the mechanism is identical across platforms. If you’re running or considering a multi-instance HA setup, I’d genuinely love to be pointed at a first real data point here. Details in the README.
Update for anyone whose device_tracker / binary_sensor entities have been flapping: arrival and departure detection has been tuned to cut down on false triggers from GPS jitter near area boundaries.
This came out of reports on the Hubitat side — users seeing presence bounce back and forth while standing near the edge of a defined area. Departures now hold for a short stability window before being reported, rather than firing on a single boundary crossing.
Trade-off worth knowing: arrivals can take up to 90 seconds to register, since it’s confirming you’ve actually settled somewhere. If that’s too slow for an automation you’re running — unlocking a door on arrival, say — let me know and I’m happy to talk through options.
One thing that matters more than I expected: area radius. A smaller area needs better GPS accuracy before a position counts as inside it, so very small areas can behave less predictably than larger ones. Still working on characterising that properly rather than guessing at a number.
No HA-specific reports of flapping yet — if you’ve seen it on your dashboard, I’d be interested to know whether this improves it for you.
