That's totally fair and reasonable. I still will demand that it be able to function in a bang-bang mode using dry contacts, especially from a new entrant to the market, because there's a fair risk that the unit will be in my house longer than your company is in business.

This behavior is how current variable speed heat pumps operate with Ecobee and Nest. You leave efficiency on the table, but I get it provides some future proofing assurance for homeowners. Definitely something we can integrate into the board.

> I get it provides some future proofing assurance

I don't think you're getting it, and a lot of us around here want you to. If I were actively researching heat pumps (a thing most homeowners do before dropping 4-5 figures on ANY hardware), if you are locking me into your ecosystem -- or even have the appearance of locking me into your ecosystem -- your product is not getting onto any short list from me or likely anyone else around here.

All of us -- every tech person who has ever gotten into home automation in a real way -- has thrown out some hardware rendered useless by a company. We've already seen this play out. You're an unknown and so are the riskiest kind of company to buy into.

You need to get this -- deeply -- if you want to sell to this market. The pull quote above makes me want to run for the hills.

Agreed. As someone in the market for a good heat pump, I'm already turned off by this offering. I don't want more features (air quality module) that are more points of failure and complexity. I just want an efficient heat pump that'll work reliably for a long time, integrate well with HA, and won't cost a kidney to install.

This all especially true given the expected time frame a heat pump is supposed to have. No one wants to spend $14,000 for a paper weight. Whether that’s because the software disappears, spare parts can’t be found, or it’s impossible to service. Any of the above are real risks that need to be considered seriously and addressed. HVAC can also become an issue when trying to sell a house, so it’s not like trying to sell to traditional early adopters. The majority of this market is going to be very conservative.

I’d honestly just try to market to heat pump believers first who actively are looking for what they are selling. First, it’s an easier sales pitch. Second, they will be more forgiving of growing pains. Go through the 1-3 years trying to get the manufacturing, maintenance, and installation locked in before trying to sell in bulk to a larger market.

As it is, it’s just too risky for many to buy. (And for my house, it would be impossible to install).

Yes, that MAY be required for full optimization

Wonderful option to have: "Subscribe, and we'll improve your efficiency by 10%++"

OPTION

Another option would be "we'll send you a download with the latest weightings for our optimizing AI firmware for $14.95, or every month for a subscription of $49/year" whatever.

I, and most other sharp people will happily ignore you and buy a system that is 50% less efficient and twice the cost to avoid your lock-in.

I'll also happily buy and subscribe to your reasonably-priced optimization service as long as it is OPTIONAL.

Just look at the nearly violent reaction that BMW got when they suggested that their heated seats would be a subscription option.

If you want your company to die before it even gets started, keep making excuses for why you need to lock us in.

If you want a growth giant, architect it from scratch so someone can happily use it where Starlink doesn't even service, and offer extra-efficiency OPTIONS that OPTIONALLY use internet service.

And yes, open-sourcing the code would be a huge step in giving people confidence that you are serious, and that their investment has a future that is not a brick.

Yeah I wish these companies would “get it” but they never do.

Provide an open API with the option to use a cloud service. Some cloud services are great, and have way better UIs than the self hosted stuff which I would gladly pay for. But take away the option to go self hosted and I lose all interest. If a device can’t be controlled locally you don’t really “own” it, you’re just renting it from someone else.

98% of consumers who would buy a heat pump have no idea what an API is.

That's quite generous to estimate that 2% of the population knows what is an API !

My first guess would be more like 99.98% don't know about APIs

But yes, that 2% or whatever is an important set of leaders

This is not strictly true.

Mr. Cool has a 2-Ton heat pump (Designed to work with existing ducting) that can operate off of a normal non-communicating thermostat. It still hits 19 SEER and does a lot of that optimization on the controller side. And when it comes to mini splits there are lots of them that will work with a normal thermostat.

I know this, well, because I’ve spent many countless hours trying to find something that will work with Z-Wave and not a proprietary communicating system. Granted it might be less efficient but efficiency is way less important than having the ability to do that. The only exception I would consider is if the communicating thermostat was fully open and had a decent API that worked without Internet access.

This doesn't seem at all worthy of anything proprietary, though -- a simple RS-485 or 10BASE-T1L protocol and a page of docs would do the trick.

And if you build this, you can also sell the same hardware plus a BacNET or MODBUS gateway for a bunch of money to the commercial building management types :)

Upvoted,

BACnet already exists as an ISO standard: https://www.iso.org/standard/37298.html https://bacnetinternational.org/

The example application in the Ethernet Alliance one-pager on single pair Ethernet is HVAC: https://ethernetalliance.org/wp-content/uploads/2020/10/EA_T...

FWIW, if this were my startup, I would probably use 10BASE-T1L. The programming model is simple, and cloud or other “smart” integration is natural if the system already uses IP. For that matter, one ought to be able to support wired 10BASE-T1L and wireless Matter/Thread thermostats without much duplicate effort.

This could be prototyped on a BeaglePlay :)

FWIW,

BACnet allows multiple PHY implementations, including 802.3

I suspect that 802.3.cg is also kosher, but haven't looked in detail. It's been 5 years since I looked seriously at whether BACnet was the correct solution for a product family.

Single pair Ethernet is indeed cool tech.