It's hard to sell OEM licenses when OEMs are not selling hardware.

It's the singular thing that got them to think about efficiency - nobody can afford enough RAM to run their shitty operating system, so nobody is buying.

Sending all your data back to themselves.

Literally the only Apple device I use is a corporate Macbook, because they don't allow Linux on the Dell hardware that is the only other option and I'm not going to saddle myself with the Windows shit show.

But you do you.

And what were you doing for that 7 minutes while it was working? Staring at it?

Guess where the productivity gain comes from? Not staring at an AI while it works. Tell it to do things, and then hide the window and do other things.

Who gives a shit if it takes seconds or minutes, if you have other out-of-band tasks to take care of that always fall off the stack because you're trying to meet a deadline or constantly being calendar-raped by meetings?

Every sufficiently advanced technology changes work patterns, and this is no different. Set the AI to working on some shit, and then do other shit.

At $200/month you may as well buy that office worker a $4000 local device with 128GB of unified RAM and sufficient compute to run a decent token rate. There's an ROI argument to be made there:

200/mo * 12 = $2400/year

$4000 / 2400 = ROI payback in 1.6 years on a device you depreciate over 4 years. And your data stays right there on the desk.

This scenario seems FAR more likely:

1. Anthropic is in the business of making VERY large models, that require a fuckton of hardware, so they have a natural competitive moat that extends beyond their proprietary code: sufficient GPUs / RAM / storage / energy / cooling / etc.
2. Their secondary competition (i.e. not OpenAI, Google, or Meta) make small models that people can run at home, largely for free, on hardware they already have, and these capabilities are increasing practically daily
3. Anthropic is burning out the clock to an IPO cash grab

Sounds to me like they are spreading FUD about small models, specifically to keep themselves relevant until they run through the tape, but I'm a cynic.

And how does the model run?

Does that camera have a shitload of RAM in it? Does it have sufficient CPU power in it's 32-bit microcontroller running in the Pentium II MHz range for an AI to even preload without shitting the bed?

So there's a bunch of slow storage. Whoop de fucking do.

This was my first thought: well if someone can get an AI model small enough to fit on an ESP32, have at it.

What horseshit FUD.

The EU doesn't give two shits if not privacy-raping them costs you money. That's the cost of doing business in the EU.

Do it, or don't do business there. Whining about it isn't going to change anything. There are zero EU politicians that are going to say "oh, we didn't think that trillion dollar companies would have to spend 1% of their revenue per year to comply with not privacy raping us, so we better just bend over and take it forever!"

What the hell were they trying to get here? Sympathy?

Laughable.

To be clear, I run Home Assistant and have integrated it with a local LLM and MCP server to allow for useful voice commands like "play the next episode of Landman in the living room" and it will look at the state of each required device and remediate them, while also searching all available services including my local Plex host for Landman, and launch the proper app with a deep-link intent.

But none of that was easy, and every time a device changes it's firmware (looking right at you, Android TV) there's a very real chance shit breaks.

When that kind of functionality "just works" is when home automation will be ready for the consumer market. Not before.

Also, this device looks like a massive Apple Watch glued to a HomePod. I don't want that.

Bad metaphor. Fyre Festival can't retroactively patch their fuckup to make it what everyone was looking for. Attendees will always have been swindled.Hello Games can.

As it turns out, software isn't linear time and meatspace event planning. Software can improve with feedback, and purchasers can get the original vision plus more. And did.

The only tech company that even remotely understands the concept of "durable goods" appears to be Tesla. And even then it's a little shaky, but they've proven themselves by continuing to update software in 10+ year old cars OTA with no bullshit.

Buy this thing if you want to replace it 5 years from now.

In a business environment sufficiently scaled to need the equipment the guy is talking about, if you don't have someone that understands TLS in 2026, you have bigger risks than anything you just mentioned.

You can trust your own certificates in an Apple Keychain. This is clearly an enterprise of some kind - if they are using Apple devices, and not using device management when Apple makes this unbelievably easy, then they get the runaround.

Nice that they support their own protocols + matter / thread; what about anything else? Is that going to be a game of "find the app to dim the lights"?

Why don't these fools ever learn from Home Assistant - build integrations, not protocols.

What consumers want: Home Assistant, without hours of setup and scripting.