Build things you want for yourself and publish them. Don't worry about cost optimization, production at scale, patents, etc.: just make something with unique functionality. Repeat. You'll be pleasantly surprised how quickly you learn and amass a portfolio of unique toys:)
Hire a PR agency halfway through your campaign when you're so overwhelmed with backers and attention that you can't handle it yourself. Until then, DIY:)
I believe we're in the earliest days: there's no dominant platform, and no 'killer use' for home automation / iot. I'd focus on your product, maybe adding support for platforms later.
Check out the common part libraries released by Octopart / Seeed , and consider getting a resistor / cap library from your neighbors at Adafruit. Hackerspaces can be good resources for spare parts (and finding potential partners).
Usually, abstractions like this serve as excuses to not launch. Launching doesn't fix everything, but it does increase your speed of learning dramatically.
See http://www.ycombinator.com/hardware for current campaigns + products actively shipping. Nebia et al aren't on the list: companies get dropped post-crowdfunding and re-added when shipping:)
Don't copy people's products. Other than that, companies spend too much time too early worrying about (and spending money on) IP. Prove demand for your product by launching; you've got a year post-disclosure to file a provisional patent.
what an interesting concept: technology as a constraint forcing you to focus. not just focus: focus more than you naturally would. focus more than you're comfortable with. continue going doing keeping on, regardless of whether you're comfortable with how it's going, with what you're keeping. will it all fade at the end? yeah, definitionally it will. fuck: and a word + character + wpm counter at the bottom? really hitting all the unhealthy adictiveness buttons. or healthily addictive in the right quantities? what's the ld50 on a writing app? on an app of any sort? the most dangerous app? i wonder if it'll be entirely different than the most-regulated app, a la chemicals. fuck is that really all i can do? 2:30 worth of content, including rambling digressions? it's a very real possibility. this fades, faster than the normal recession to black/white.
...
let's see, what else is on my mind...
...
to stay or go? i failed at ignoring politics, in spite of my success since protesting in 'free speech zones' at bush's reinauguration rnc in nyc. (found out years later that many other riders of critical mass got arrested) i'm convinced trump is going to win. it's bad: he's the president the average american deserves. the world doesn't, though; that's nearly narendra-modi-armani-level sad. and likely more negative at scale.
It's about optionality: startups of all types can now do more with less. SBIRs, access to college labs, in-house prototyping facilities: these can certainly help, but what's exciting is that they're no longer required to iterate on prototypes. So little funding and time is that it looks a lot like software.
While I have no idea how many you specifically will choose to purchase, a majority of the hardware startups in our current batch will have product for sale via Amazon and other channels by summer 2016. This sounds strong given the history of hardware companies, but our current batch is not comprised of average hardware companies;)