HA is my hobby so no, I’m not an installer.
For basic things RPi+SSD should do but it’s much less powerful than NUC/PC.
Again, I’m looking at it from a hobbyist’s point of view, your approach should be focused on reliability/maintenance cost etc.
p.s you can Edit your posts to add something rather than post each bit separately 
I run a Pi 4 with SSD (Not particularly easy but it’ll get there) Pi 3’s with SSD’s are VERY Easy
I highly recommend starting with your own home and test if for at least 1 year and go with all releases, through all breaking changes and after 1 year think again whether you want to do all this for clients as well.
The only business model I see here which makes sense would be a subscription service. A flat monthly fee to do upgrades and minor customer-requested changes. Using remote access it shouldn’t take all that long, but it would be a nearly daily chore.
Also consider obsolescence. As was pointed out above, just about all the hardware will be old and in some cases unsupported after a year or two. You can spend a LOT of money putting in a top-notch system worthy of your reputation, only to have it look quaint, or not work at all, very quickly.
Well, how often a manufacturer of your car updates its firmware? Same about your HDD/SDD?
I don’t think it’s necessary to keep customers’ systems constantly updated (and there will be much less important ones for a stable product for sure). So stable version + good system hardware and reliable periphery should live longer than 1 year.
PLAN
Answer below:
1.what would you like to have?
2.what do you currently have?
-how this differ from #1
-what works well now
-what not work well
3.who will use?
-what are their needs
-what should they not access
-what technical level
Answering above can help when choosing product and design. For example:
If for Airbnb house do you really want interface to by app and force download. Web is universal for this case and less hassle for user.
DESIGN
What products, what technologies and how connecting/installing?
Do You use 433mhz, zigbee, wifi or zwave?
-wifi device usually mean app needed. Company only support product/app maybe 5yr. $20 echo…who care but $250 Sonos bad deal. Consider how support will work and cost impact when design.
Interface by app, tablet, phone, PC, what?
What service level needed?
-if HA go down, what is impact?
-what is Backup when go down?
-design to meet these requirements
I think you get general idea.
Not true
(2)yrs outdoor in poor environment with cheap psu. No problems. I expect it to break tommorow however since I posted this 
That said, I disabled logging in this unit to prevent/reduce writes to SD.
If you are looking for a reliable wired setup without proprietary risks (modules no longer available for example) an industrial PLC is the only and best option to my opinion.
PLC hardware easily lasts for 20-25 years and spare parts are produced for a long time (check out the Wago 750 series if you don’t believe me).
For integrating the PLC with HA have a look at the below:
There’s already a few people using that software to control their (wired) homes with a PLC and neither the software or the PLC has ever failed .
Great thing about it as well is that all critical tasks like switching lights, controlling covers etc are perfomed inside the PLC software (but can ofcourse still be controlled otherwise). So if you screw up with HA (perhaps during an upgrade) or your HA hardware fails basic things like lights, covers etc will still work…
A PLC comes at a price but in regard to reliability it is definitely worth it.
Only ‘downside’ is that is a wired solution and thus can only be applied on homes that allow control through wires with a central system to control it all…
Personally I think that wired means more robuust but that’s just my opinion 
Good point but imho it’s more suitable for a person willing to make their house reliable and future-proof.
For an installer it’s a bit too much both in terms of cost (less clients) and longevity (20 years is too much).
This is not exactly true
A light switch is light switch and no amount of technology advance will change that.
Some zwave switches haven’t changed in years and honestly they’ll likely be OK 20yrs from now.
Wifi switches are kinda the same when on vlan but their apps will die
Colored LED lights are a trend and will go the route of shag carpeting at some point.
LED bulbs on the other hand are functional and easily replaced when they die or better option is presented.
Echo, Sonos, Google home are trend items that eventually will be replaced.
An amp and speaker can last the life of the home. Paired with an echo or Chromecast you have 2020 audio system and in 10 yrs you can add the next trend device to have a 2030 audio system.
My point is don’t use trend products in structural/necessity functions. Identify trends vs long term and you can avoid the setup that functions like the robot from Lost in Space
Michvw, that is super interesting !
Could you kindly give me references to industrial PLCs that are reasonably priced ?
This would definitely be the way to go for wired connections. Do these central PLC units have a fail-safe system ? If something does go wrong with the PLC it could cripple the lights in the entire house. For lights , it would be good to have a hybrid system that defaults to manual should anything happen to the PLC unit.
With retro-fit installations, I suppose each Z-Wave / Zigbee switch becomes a stand-alone controller for its associated circuit. In such a case, if the associated mesh-network goes down then one would lose control. However, even if the mesh-network or WiFi infrastructure goes down, manual operation directly from a light switch should continue to work.
Furthermore, clients like to see security cam feeds on their smart phones or domestic tablets, this is bandwidth intensive. I would recommend industrial strength managed switches and wired 802.11AC APs. Not sure about the AX standard, is it much better / faster ?
I know self diagnosis and self healing systems will eventually become standard features. Smart homes will no longer be a luxury, they will be a norm.
Congratulations and nice work Browet. However, I feel the interface could be better. I guess it all bottles down to personal taste and preference.
If I may ask, what sort of switches are you using for the boilers ?
Good points - Reliability remains high on my list. Most of my clients are people that are not technical at all and would be happy to learn and use the system. Hence, a nice UI and a pleasant UX is key. Rarely will they want to make a change but they will want the system to work efficiently. A change would be rare and would entail adding a device such as a new lamp in the lounge or a timing schedule for a hot-water boiler.
I would disable auto updates to avoid putting an operational system at risk. I know commercial providers work very closely with key device partners to ensure a smooth problem free upgrade transition. Hence, the wisest thing for an installer to do is stick within the recommended list of devices for stuff like light switches, cameras, door-locks and shades.
The situation becomes a little tricky for specialistic AV gear but reputable commercial providers are able to write drivers for devices that are not listed in their database.
I’m new to all this stuff but I am trying to research the subject carefully so that I can eventually make an informed decision.
I must admit, members of this forum are awesome.
Thank you and the interface is “work in progress” since the beginning…
For the boilers and big power consumption equipments, I am using TP-Link HS110 (with power usage capability)… HS100 for small equipments (lights for example). I am also using some Sonoff (R2 with 4 channels) for the pools and remote location … or even other brands found on Amazon (some I had to flash them as they were not recognized by HA)… Cheap but not necessarily easy to integrate…
I’m sure these in-between sockets are useful for plug-able devices but I would need a switch that is installed into the wall and can be activated manually or via the control system. It would need to be 30-Amp capable just to be safe. Once it is turned on, it should turn-off automatically after 30 minutes. Early morning and evening schedules would be nice. Most clients forget to switch it off and realise after a few hours, this reflects in their bills.
For 30 amps, I would use a “cheap” switch (with capability of communicating its state to HA: “on” or “off”) piloting a relay to provide the power you need… 30 amp switch can be very very expensive if exists… On the other hand, HA is able to determine if an entity (a switch) is “on” for x minutes (if the state is “communicated” to HA) and HA can take action if needed… So you can take action based on its uptime and/or the hour of the day and/or special events like sunset and/or even based on email content… Very very flexible, possibilities are unlimited… The limit is just your … imagination… 
Aeotec used to do (prolly still do) a 40 amp switch. The point is what do people mean by “feedback”.?
Any contactor could fuse closed or burn the contacts to make it “not close” (though that’s much rarer)
So buy a cheap Wall switch of whatever type you favour, put it in a Box with a contactor of what size you need (40, 60, 80 100 amps) and you have a *#%k£^% big switch (just make sure the box can breathe and mount it properly)
Basically I think I’m just agreeing with you 
That is both a strength and a weakness IMHO. Jack of all trades and master of none.
If home assistant had started as concentrating on one technology “we’ll be the best wifi/zwave/zigbee (pick one, or any other) home automation system” it’d be at version 3.0 by now and a stable commercially viable system. But it wouldn’t be home assistant as we know it.
Right now it is the most powerful and flexible systems I have seen, but it requires work of a sort that a custom commercial installer is going to struggle to make money with.
If I was to use it on that sort of basis, I would choose a very short list of well supported components and stick with them. I’d lock the installs down so no one can fiddle (this seems no different to control4/crestron etc) and make up for the fiddling required by charging ultimately what the pro systems cost.
Thank you Browet and Mutt - The relay approach makes sense and would probably be much more cost effective. Aeotec does a 20Amp switch, will check out Prolly too.
Hi Nick, that is exactly what I have in mind, to work in a highly systematic matter. As tmjpugh suggested, it is important to PLAN and DESIGN carefully.
As per your recommendation, I would strictly work with a short list of well supported devices. Device manufacturers that recognise and support HA. My clients will never insist on certain brands and I will not go into an existing installation in order to improve it or expand it. 99.9% of my clients are interested in home automation from scratch and rather have me suggest and incorporate an automation infrastructure. I have 2 types of clients: those that already have houses and will not want to go into the cost and complexity of wiring and those who are building a new house and willing to wire everything for the sake of stability. My challenge is to make retro-fit installations super reliable as well.
The overall idea being is to minimise service calls pertaining to problems. Service calls should pertain to expansion because clients are loving their systems.
Is there a list of such devices, especially light switches/dimmers and cameras ?
Has anyone successfully managed whole house music system that is as good or better than Sonos ? What I like about Sonos is it’s ability to rely on Sonosnet mesh network that is independent of the main domestic WiFi network and one can use Boosters to cover dead areas. The fact that the speakers are active is nice and I love the fact that subwoofers can be integrated easily and they have decent performance too. Best of all, the reliability factor is high and their support is second to none.
Is the interface design aspect of HA easy and flexible? The one indicated on the HA site is not bad at all. The best I’ve seen is Crestron Home, very clean and intuitive (Apple like).
Still waiting for Michvw to point me in the right direction in terms of industrial PLCs or an ideal PLC. I too believe that wired is the best route to take and I’d definitely prefer running CAT-6 than a slew of cables including speaker cable and coax.
What about AV control for entertainment areas and dedicated home theatre rooms ?
Nearly forgot - Is Pi4 with the SSD hack the best way forward or is there an alternative that is better in terms of reliability ?
HA should introduce an Pro version (HA Pro) for installers where installers would be able to buy the Pro version for a reasonable price. This would fund a small but highly efficient development and support team specifically for HA Pro. Money earned could be used to better the HA Core so that it remains open and non-profit. This would benefit the DIY community, installers and end-users. Furthermore, it would allow everyone to enjoy the benefits of home automation due it’s power, flexibility, reliability, affordability and excellent UI/UX. An innovative and revolutionary approach.