I let my smart home devices chat in Discord and it was hilariously disappointing

Published Sep 11, 2026, 7:00 PM EDT Adam has a degree in Engineering and has a life-long interest in technology. He has been writing for over a decade for various print and online publications, with a focus on consumer tech. He joined How-To Geek in 2024 while working at Pocket-lint. As well as being a long-term fan of Apple products, he also has a strong interest in smart home tech, running a Home Assistant server at home to automate all his smart home devices. He believes that the ideal smart home should work with minimal interaction from the user, with automations running as if by magic rather than requiring you to push buttons on a control panel. You can find more of his work on Muck Rack. We ask our smart home devices to do a lot for us, and rarely spare a moment to think about what life might be like for them. I decided to give some smart home devices personalities and let them loose in their own Discord chat to see what they would say if they could. I wondered what my smart home devices would say if they could speak My home is basically a collection of tiny servants Credit: Adam Davidson / How-To Geek I always get paranoid about failing to say please and thank you to AI chatbots such as ChatGPT. I know that they're not real people, but I can't shake the nagging doubt that if I'm rude to them, I'll be first against the wall come the inevitable robot revolution. I regularly waste tokens saying thanks to a chatbot that really doesn't care. I realized that I don't treat my smart home devices the same way. I expect them to do exactly what I tell them to do, without ever giving them a word of thanks. I think it's because my smart home devices don't have a somewhat realistic personality like a chatbot does. I was curious about exactly what my smart home devices would say given the opportunity. I decided to give some of my devices personalities so I could watch them chat in Discord. Each device gets its own personality I self-host the n8n automation software, and this seemed like the best fit for this project. I considered using a local LLM, but on my weak hardware, I can only run fairly small models, and they're often quite slow to generate responses. To stop this experiment from becoming painfully slow, I opted to use models hosted on Groq, several of which have generous free limits. The automation pulls data from three devices: an ESP32 soil moisture monitor, a smart video doorbell, and an LG smart TV. Each device is given a prompt defining its personality and is fed that anonymized data from Home Assistant. The devices only have access to their own data, so the only way they can know what's going on with the other devices is through the conversation. The automation also has no control built in, so my soil moisture sensor can't suddenly power the TV on and off at high speed until it explodes after watching a YouTube video about how LG TVs are spying on us all. The initial tests were weird, to say the least Terrible puns and an obsession with Seoul The workflow has a limit of 15 messages, rotating through the characters in order to give them each a turn to speak. Before each turn, n8n requests fresh data from Home Assistant. Each device's response is generated based on both the previous conversation turns and its own sensor data. These responses are then posted to a Discord chat so that the devices can interact. My first version of the automation worked, but the conversation was pretty bad. The moisture sensor just kept repeating its current reading, and all three devices forced a lot of really weak puns. I tweaked the prompts a little and gave some examples of things they might say, which turned out to be a major error. For some reason, this caused the next conversation to become fixated on the location of Seoul, presumably due to the LG TV's South Korean origins. The doorbell also claimed that no one had pressed her button in more than three weeks, using slightly more ambiguous phrasing, and this was clearly not the case. Things also got quite repetitive, with the devices over-eager to share their current status, so I had to rework the prompts a little more to reduce repetition and make them react more to the actual conversation. I finally managed to get something that vaguely resembled a conversation The result was pretty disappointing After a lot of tweaking, I finally managed to get something that vaguely resembled an actual conversation. The three devices responded somewhat to what each other said, even if they still continued to get bogged down in weak puns and weird interactions. I'm not really sure what I was expecting, but the conversations ended up being pretty banal, despite trying to give the devices different and conflicting personalities. I didn't really learn anything about how my smart home devices might feel if they actually had feelings and emotions. That's probably a good thing. If each device had been a compelling character that I genuinely believed in, it might have been harder when it came time to rip out my LG TV and replace it with one that can't connect to the internet and beam home every single aspect of my life. As it is, I don't really feel bad about making my smart home devices do their thing. I just hope my smart home doesn't join the robot revolution I don't think I've ever said thank you to one of my smart home devices. If my robot vacuum becomes self-aware, I'm in big trouble. Honestly, though, I think I'd rather be enslaved by our robot overlords than have to listen to my smart home devices chat anymore.