I built a window vent with a variable-speed computer fan and an esp8266 that communicates with a CO2 monitor and raises the speed when the CO2 gets too high. I have a very fine mesh on the window screen that's designed to filter out pollen and dust. It works really well. I usually only turn it on at night, because that's when my wife and I are both in the room and the CO2 can really build up.
I noticed this a few years ago and decided to start giving any automated systems a special email address so I can filter automated email into a separate folder. I only give out my personal email address to actual humans. Its been a huge improvement to only see human-written emails in my inbox.
I knew npm was a train wreck when I first used it years ago and it pulled in literally hundreds of dependencies for a simple app. I avoid anything that uses it like the plague.
In my experience, this is how it goes whenever there are more than 2 people in a conversation. The speed of the topic switching defaults to the fastest processor in the group, basically excluding any slow processors from the conversation.
Before I started self-hosting my LLMs with Ollama, I imagined that they required a ton of energy to operate. I was amazed at how quickly my local LLM operates with a relatively inexpensive GeForce RTX 4060 with 8GB VRAM and an 8b model. The 8b model isn't as smart as the hosted 70b models I've used, but it's still surprisingly useful.
I thought they were overpriced until I looked into building one myself. The heat sinks were going to be really expensive. The LEDs and lenses weren't cheap either.
I use Blender for CAD. It's not designed for that, but I've developed a particular workflow that fits my needs quite well. I use it to design everything from small parts to buildings. I've used FreeCAD, but it's just too time-consuming.