-rick-

Careful where you might end up...

Everything I'm seeing suggests that the world has changed somewhat over the last 4 months or so, so if you haven't tried with codex 5.3/opus 4.6 then you might have the wrong mental model of where things are.

Anthropic just announced a big increase in number of tokens included in their subscriptions. OpenAIs codex is already generous (apparently). So maybe worth another look. Seen plenty of people talking about techniques for minimizing token usage. This is the thing that confuses me the most about your plan (not that you have told us much detail). Home Assistant can already do this can't it? (some features need a live cloud connection but some don't IIRC). Even if the operational side of your plan needs a local LLM, the coding/development aspect is one of the hardest aspects of LLM use that the frontier providers are better at. So your best value for money might be using them to do the development and then using a smaller, less memory hungry local llm to do the basic daily interactions. (Should be possible to design it so it can call out to a bigger subscription model for 'hard' problems). Having said all that, if you are enjoying the process, have the spare money for the hardware then do what makes you happy.

Congrats. Work smarter not harder! Always surprised when I see people manual handling relatively large amounts of stuff when there is a machine that can be easily hired/bought and can do the job in a fraction of the time. Sure, money is a factor but in so many cases using the machine works out cheaper even if you don't value your time/health.

Given the craziness around AI it wouldn't surprise me if someone bought this, especially if the new M5 models are delayed/limited availability.

Text based communication makes it harder to get intent across. In this case I'd say your post did not land well. Having the first response to a 'help i may be in a sticky situation' post be 'well if you'd had made different decisions 2-3 years ago you wouldn't be in this sticky situaion' is particularly unhelpful.

It's related to gaps surrounding the module. The regs tightened up the rules around what you can fit in a domestic CU and the effect is that most sparks won't want to fit anything that isn't certified by the manufacturer as compatible/suitable for use with that specific CU.

Then you'd need to either use a matching brand/spec transfer switch that's certified for the same CU or have a sparky that's willing the do the risk assessment and put his/her name to the safety of the third party switch. I'd guess a separate enclosure would be easier. Once you do that, there are multiple options of pre-packaged solutions available. More expensive sure, but all-in-one and likely easy to get the sparky comfortable with. https://www.cef.co.uk/catalogue/products/1780374-63a-2p-surface-mounting-3-position-changeover-switch-ip65

Was thinking far more of guests arriving and wanting to jump in the shower. Agree A2A is fine if the room is cold/hot on arrival.

There are plenty of integrated combi microwaves that don't come with a plug/need a 16A circuit. Just seems a bit odd to need the circuit for a much lower draw pure microwave, but not need it for a high draw integrated washer dryer or similar. I wonder if it's because you work in a lot of higher end homes where the microwave is usually an oven as well and so it's become standard practice amongst your sparks to do that rather than regs?

😆 Edition (of the regs). I'm curious of the reason too. Built in microwaves are more likely to be combination units (with oven as well) and microwaves are also relatively inefficient so they consume a lot more than the output power rating might suggest. But still, if it's not a combi then its continuous draw shouldn't be much different than a powerful kettle. (I can't think of a use case for a microwave that demands full power for more than a few minutes).

You obviously need a local override. Less smart options exist (keycard switch that only provides power when the room is occupied, or 24hr timers that need activating on first occupation but both of these may lead to no hot water on first occupation). Depends on occupancy. If you manage to be near fully booked then sure. If it's intermittent then could be quite expensive for uninsulated huts.

Agree with all the above. A2A + immersion. But also, make sure you put in some easy way to only have the heating active when let (but also that the cabin is warm/hot water available when the next customer arrives). Maybe the whole cabin supplied through a contactor that has some smart controller on it then sync that controller up to your bookings system. (Sounds complicated but probably not if you use something like https://ifttt.com/)

We do have rules about firing workers to replace them with cheaper ones. Things like TUPE. But our current breed of politicians tend to focus on the wrong things and AI regulation has been low on their list. Especially since Trump got elected as they seem particularly wary of doing anything that might upset him and the tech industry makes sure he knows about anything they don't like coming from us.

MacOS will use less than 4GB IIRC. But decent LLMs are very memory hungry so I think @Pocster was more focussed on the RAM needed by the LLM not the OS. The smallest 'useful' models might be 16GB and then you need a ton more for context windows + LV cache. These models will still be far behind anything from the main commercial providers. More memory = better model + more context.