Unfortunately, there's no good API (and your choice of three variously lousy ones [1]), and it syncs in the same way as the Beddit does, i.e., via Bluetooth to a phone. (Or via USB to a computer, but that's not much more help.)
I've thought about trying to MITM the data on its way out from the PC to Basis's sync endpoint, in order to see whether I can trap it there instead of having to query it back out of one of Basis's various APIs once it's synced. On the other hand, I've already got > 1 month of data synced, so I'm going to need some method of extracting data from their backend in any case. (But on the third hand, since Intel bought Basis and Basis apparently doesn't bother much with new development any more, I figure it might be handy to have a backend for sync data in case the hardware becomes otherwise useless.)
[1] Two equally undocumented and unstable not-really-supposed-to-be-public APIs, for which various clients exist on Github in various states of disrepair, and a third, also undocumented but probably more stable, API which feeds their web UI..
I am using Fitbit and Basis is what I really need. I think by next year we will have something along the lines that Basis promised.
And I can't really knock Basis on the hardware score; the device itself is actually quite nice, from build quality to resilience to battery life. Their web UI's not bad, too, but what I really need is an API to pull that data out and integrate it with everything else I'm logging, and it just doesn't seem like they give a damn about publishing something stable, nor as far as I can tell have they ever.
On the whole, I'm pretty equivocal. On the one hand, the Basis device is excellent, and while I can't speak to accuracy, it's quite precise. On the other, the software support isn't sufficient, and the organization doesn't give any indication of being motivated to fix that. I'm not sorry I spent $200 on a Basis band, but you might be; think it over carefully.
I was hoping the Melon would actually ship, but it looks non-wearable at night.
Beddit is kind of a pain because launching the app on my phone is a pain; it requires manually pairing each time.
I'd really like a network-connected, adaptive-scheduling clock next to my bed, with EEG input. Something which could wake me up early if traffic to work is bad, or let me sleep late if my flight is delayed. Ideally with a home version which does NOT use my cellphone, and a travel version which is compatible and uses my phone.
I'd be fine paying $500 for this. There is probably a market in the tens of thousands.
1. You need the app running to collect metrics from the device (so, still some friction). I forget the app all the time; at the end of the day, I drop my phone on a charger and crawl into bed. Relying on humans to actively intervene is, unfortunately, suboptimal.
2. I was hoping it'd attach to my wifi and dump metrics to an API I could query (there's no smart alarm, so attaching it to my own stack of stuff seemed cool). Unfortunately, it sends data via a private Bluetooth protocol to your phone, rather than the wifi. Intercepting this is non-trivial (although the Android Bluetooth debugging stack helps). I'm trying to build a receiver on the Pi currently.
3. The API still doesn't really exist.
My use case is slightly different from others. I've got a chronic condition, and I'm not really interested in "did I sleep well last night?", which Beddit seems to have targeted. I'm much more interested in trends over a period of time, once my illness flares -- "am I waking up more often?", "how much time am I spending in bed, rather than active?", "over the past week, how many times have I gotten up -- should I see a doctor?". This should correlate with other smart devices (scales -- "how much weight have I lost?"; fitbit -- "am I still relatively active?") to give me a more holistic view of my health. So, long-term data retention is important to me (CockroachDB looks quite neat!).
Smart alarms and overnight statistics are interesting, but I hope companies developing devices for the quantified self start to pay more attention to long-term health data. It paints a far more interesting story :).