Twice in one week we turned the truck around at the trailhead and drove back to the campground. Both times the dog was asleep with the air conditioning running exactly as we had left it, and both times we had spent the drive certain it was not.
Nothing had gone wrong on either trip, which is what made it worth solving. From a trailhead, a rig that is fine and a rig whose pedestal breaker tripped an hour ago look identical, so the only way to know was to go and look, and going to look costs the best part of the day. Most of what follows came out of those drives, including two or three decisions that look from outside like features we forgot to build.
The sensors were already in the rig
There is an obvious version of this business and it is a closed one: our hub, our sensors, our prices, nothing else allowed in. That is a perfectly good business and it is not the one we are running, because most people who want this already own half of it. A Govee puck in the bedroom, a Mopeka on the propane tank, a Victron shunt fitted last spring, all of them already throwing good readings into the air for anything that cares to listen. Telling that owner to bin the lot and buy ours is a worse product at a higher price.
So the hub listens to what is already there. What it can hear is written down in the compatibility database, which is less a marketing page than a build artefact: it is generated from the decoding rules the software ships, so a device reads as supported when the code decodes it today. We cannot talk our way onto that list, which is inconvenient about once a month and worth keeping anyway.
Most of the time it should have nothing to say
A wall of live gauges is easy to build, easy to demonstrate, and quietly useless, because nobody really wants to watch a graph of their fridge. What is worth paying for is silence, broken once, at the moment something needs you, which puts the interesting engineering nowhere near the readings and squarely into thresholds, cooldowns, and the unglamorous business of not waking you at three in the morning because a sensor blipped.
That last part decides whether any of this is useful. A sensor at the far end of a rig drops a broadcast now and then, which is what radio does around metal and microwaves. Treat every dropped packet as a sensor going offline and the owner mutes the whole thing inside a week, and a muted system is worse than none, because you carry on believing you are covered. A gap of a few minutes is radio noise; a gap of an hour is a sensor that has genuinely gone, and only the second one earns an interruption.
The two situations we built it around
Both of these are ours rather than a market study, and most of what the hub does is a variation on one of them.
The dog at floor level
Interior temperature taken low down in the living area, where the animal actually settles, rather than at the thermostat, which sits high on an interior wall in the coolest part of the rig and will report 72 while the floor is doing something warmer. Shore power matters more than either. Air conditioning almost never fails of its own accord: the pedestal breaker trips or the loop loses power, and the rig starts climbing from a number that looked fine when you locked the door. Reading the supply tells you that at the moment it happens, rather than once the heat has caught up. The pets guide covers the placement and the threshold.
The rig in January
Off season it sits somewhere we are not, and two questions are worth answering from the sofa: whether it is getting cold enough to do damage, and whether something in there is still drawing power that should have been switched off in October. The second is more interesting than the first, and it has its own post.
Where it stops
A monitoring product drifts towards life safety if you let it, one sensible request at a time. At the end of that road is a phone notification standing where a certified alarm should be.
This is not a life-safety device. RoamVitals is a monitoring and convenience product. It does not replace a smoke alarm, a carbon monoxide alarm or an LP gas detector, and we deliberately support no device in any of those three categories. That is a decision rather than a gap in the roadmap, and the reasoning is here.
It is not a security system either. Put a sensor on a door and it will tell you the door opened; it will not tell you who, and it is not going to ring anybody on your behalf. It also depends on an internet connection. The hub rides out short gaps and catches up afterwards, but a rig parked where nothing has signal is a rig we can tell you nothing about, and we would rather write that here than sell you certainty we cannot deliver.
What works today, and what waits for launch
We launch in Q4 2026. Email, SMS and Pushover alerts are built and running in our own testing, and they go live for customers at launch alongside the hardware, in the same words we use on the pricing page. Alert delivery is the whole reason to buy this, so something that works later has to be labelled that way in both places.
None of it makes the drive back unnecessary, and some afternoons you will turn around anyway, which is fine, because a quiet phone is evidence rather than a promise. What it changes is how many of those drives you make for nothing, which for us was very nearly all of them. If there is something you want it to do, the request board is read by the people writing the decoders.
