First, you should read this this question here.
The reasonable answer here varies a lot depending on how involved you want to get, and what volumes you want to sell at.
Fortunately, you're not using any breakouts with truly exotic parts. The nRF8001 is just a QFN part, which can be soldered by hand without too much trouble.
Speaking loosely, the process would look like this:
- Design a PCB layout that contains everything.
- (Optional) - Figure out if you want to put it in a enclosure of some sort.
- (Optional) - Determine how you're going to have the enclosures produced or modified from an existing product. You can have a contract machine-shop do the work, but again, $$$$.
- Send out to have PCB prototypes manufactured. I like OSHPark, but there are LOTS of options.
- Assemble your boards.
- Find out your prototype doesn't work.
- Go back to step 1 (expect this to happen a few times)
- Rent/Beg/Borrow some equipment to do EMC pre-qualification testing (for FCC).
- Realize you have an EMC disaster.
- Respin boards a few more times (e.g. go back to 1 again) to fix the pre-qualification EMC issues.
- Feel like you have a decent layout, have it FCC tested.
- Pay $10K+ for FCC testing.
- If you're lucky, you passed, you now have a proper, FCC certified product.
- Nope, go back to 1 again if you failed.
Do lot assembly, either manually (major time investment), or contract out (expensive!).
Sell your widget!
Note that you can contract out the prototype assembly, but doing so is expensive, even if you do it in china. The magic (or cursed, really) term here is "NRE", or non-reoccurring expenses/engineering. Colloquially, it's the "setup cost" of doing a production run. Sure, you may only pay $5-50 per-widget to have them assembled, but there will be a $1000+ base charge. This reflects the fact that the primary time-cost on the part of an assembly house to build your widget is setting up the production line, working up and debugging a pick-and-place routine, building test-setups, etc... Once everything is up and running, the actual per-unit cost is fairly marginal.
This is also why commercial electronic devices can be so cheap. They're made in truly enormous volumes, and that lets the companies amortize the NREs across the entire production run. NREs are basically the bane of any small electronics business. They're also basically the reason specialty widgets are so expensive. Places like spark-fun can get away with fairly cheap products because they have their own complete assembly line in-house, and even there, you're often paying 4X+ the cost of parts for just the IC on a PCB with a few cents worth of passives.
Since you seem to be new to PCB design, this is possibly not a great first project. It's doable, but expect it to cost 4X+ what you expect, and take 2X+ the time you planned. There are a LOT of facets to designing a proper product, and they each contain an entire field.
It'd probably be better off to start out with some simpler boards, maybe a through-hole-only arduino clone, or similar (though SOIC can be soldered without a hot-air station or anything).
Some assumptions I'm making:
- If you're serious, you will have to lean SMT assembly practices. Through-hole only is not a viable stance anymore.
- Expect to buy a lot of tools, and burn a lot of money in failed prototypes.
- Expect 3 board revisions. At minimum, before you have something you can get tested. I could maybe do it in 2, if I got really, really lucky, but I doubt that would happen, and this sort of thing is my day-job.
Hardware is HARD, and worse, it's completely unforgiving. If you're coming from a software engineering background, you basically have to discard all your assumptions, as they will actively hurt you.
Imagine a software project where "compiling" can take 4 weeks. That's hardware development.