Thanks for the link. Given the proliferation of cheap cloud-based Tuya and Tuya-like devices, I’m not entirely sure I agree with its premise.
I thought this part was interesting:
I think the biggest issue with someone transitioning from a cloud-setup to a fully-local setup is the cost of replacing a whole bunch of cloud-dependent devices. I think Matter is making that easier.
Makes me happy I found Hubitat before jumping down the rabbit hole.
Ditto. And add HA
I think the issue is HA in general is not for the average home owner. You have to have a technology bent and be slightly to moderately technical. That excludes 80-90% of the market so right off the bat, leaving only a sliver of the available households.
Until systems become more user friendly probably via AI front ends to hide behind the curtains drivers, apps, scripts, scenes, and rules I don't see much changing. It just too steep a hill to climb for non technical people.
The flip side is the platform that solves this puzzle will likely change the HA landscape bu opening it up to more of a mass market.
I think WiFi routers have done better here than HA. They have simple nominal setup modes for the 90% use case and then expert modes for those bent on customizing their network.
I think this is true for just about everything - from phones to tablets to PC's and MAC's.
While most can make calls and perform basic functions, there aren't many who actually know how to use them and can do everything they really want to do. As one who plays tech support for most of my friends and family, I see the logic gap in a lot of minds who just don't understand how their tech works and blame the tech for things they break. Until devices can work around those, it doesn't matter what system they are on, someone is going to have to be able to fix things when they don't work as intended.
Note: Apple tries to do this, but the cost is restrictions that hobble the power user and I still play tech support to my Apple friends who can't get things working - even as user friendly as it is designed to be.
Adding AI (as the article suggests) is likely to make things worse until such time as AI is more evolved and reliable.
I agree, then again, not much of anything local is.
I worked for an IT distributor (Fortune 100 then, Fortune 50 now) when the IoT fad hit. I was assigned to the IoT division and tasked with establishing our presence in the emergent market.
I was the only engineer on the team and was tasked with making sense of the hundreds of vendors we rapidly onboarded.
After a year of thrashing hardware / software solutions. We came to the conclusion that as an IT company our best play was to focus on the analytics. We kept a handful of vendors that were profitable and focused on what we and our partners did best; sell IT equipment through our reseller partners. We were not the only ones, our competitors followed suite.
The money that drove us came from large IT players like Dell, HP, Cisco, and every other wannabe tech giant, dried up. A large portion of that spend was advertising; the more a publication writes on a topic, the more advertisers will spend with those publications. No IoT ads, no IoT articles. Smaller players that were riding the wave no longer benefited from free market momentum.
To bring this back to home automation; the rest came down to simple competition, perfectly good products like Wink and Iris fell to the more competitive products like Hubitat.
That's my perspective on the matter. Now maybe I should read the article.
Michael L Nelson, author of the Return from Cataclysm Saga.
Speaking of which, my local voice assistant implementation is going great! I can control almost all Hubitat devices by speaking into the Home Assistant Voice Preview Edition devices and using the HA/Hubitat integration. The LLM doing the thinking part is currently mistral-small:22b with Ollama running on my 2018 vintage NVIDIA 2070 8GB GPU. It can even give decent answers to questions like "What are the benefits of honesty?" and "Which is better for cooking, stainless steel or aluminum?" and "What is a good temperature to make garlic bread in the oven?" - all local. It is not nearly as good as Alexa for voice recognition but it's getting better. WAF rating still unacceptable. I now have about 20 CD's from my ancient collection living in Jellyfin. This weekend ChatGPT is going to help me link Jellyfin and HA/Hubitat with voice recognition and AI so I can say "OK, Nabu - dim the lights in the living room and play some James Taylor from my CD's on Jellyfin." Watch out, Alexa! We return you now to the topic already in progress.
I should add that I repurposed the Raspberry Pi 4 that was running my Echo Speaks Amazon token server with a HiFiBerry clone optical digital audio hat from AliExpress to be an endpoint for Jellyfin sending FLAC audio by WiFi. The digital audio goes directly into the optical inputs on an old Yamaha receiver and out to even older Altec Valencia speakers without MP3 or Bluetooth degradation. No James Taylor on the tiny speaker in the Home Assistant Voice Preview Edition boxes, please.
For me, while I do think the complexity of any likely setup, cloud or local, can be a steep learning curve for many non-tech-savvy users to navigate, I still feel that the logic / automation aspect is still a more significant hurdle to overcome. Many people can likely draw on experience they have had with similar technologies to start to understand how smart devices fit together, but the ability for people to grasp both the logic required for an automation, plus the technical intricacies of the devices involved and how these interact with the automation, that can be a stretch for many at the moment, IMHO.
So if a significant barrier is the human-to-machine interaction, i.e. configuring an automation, then it does feel like AI could play a role here. The ability for AI agents to take a conversation and translate that into something more technical, including refinements that can be made in an equally conversational interface, does, to me, feel like a likely way that people could start to use smart-tech more.