Agents implemented 6 ways to update a smart device, on real hardware
AI coding agents implemented six ways to update a Matter device on real hardware.
A random mix from the whole library. Shuffle for a new set.
AI coding agents implemented six ways to update a Matter device on real hardware.
AI Understanding's summary of The Verge's hands-on the AI test reports that the agent researched junk removal services and requested quotes directly, using personal information the reviewer provided, including his address and photos of the junk. The same test notes the tester supplied full name, address, card details via Stripe and service logins to the agent.
He had wanted a small Home Assistant remote on an old M5Paper eInk module for 3 years and never built it. The AI did it in half an hour, and the remote now pulls up all his rooms and shows the sensors and buttons he tagged as Remote.
Vancouver Tesla owner Devin Olsen used AI in his car to order Starbucks while Full Self-Driving drove, confirming the drink and total by voice before the AI placed the order for pickup on arrival. He described it as the first time the car felt like it could run an errand with him.
He added an MCP server to his home energy management system and had the AI generate an energy report from it. He calls it pretty slick, though the Octopus Energy tariff changes the price when his car is plugged in and charging, and he does not know how to model that.
Spent an hour with AI migrating a Home Assistant setup to Tailscale, updating sensors and Pi-hole, and putting it all in one documented git compose file.
The maker said his new Leica DISTO D2 laser measure did not work with Linux. He and the AI worked out how to connect it to his Linux computer anyway, so he can capture data for his new apartment.
Working from a machine an agent had never run well, the AI set up his home server stack over Telegram voice messages: SSH, Docker, Traefik, certificates, split horizon DNS and local models. It also runs his house watching through Frigate and a solar dashboard he can check from his phone.
Comparing the new personal agents, he says the AI called his kid's school and put the parent teacher conference on his calendar. He still rates the AI ahead of the AI Duet for his own use.
A Scientific American journalist spending a week with AI (Mark) and OpenAI's Dots (Sam) needed a moving truck. Her own search found no 10-foot trucks at the nearby U-Haul lot. Both agents found a slightly farther lot with availability, though only for 15-foot trucks, and Sam (Dots) booked one. At the lot next morning, a cashier mentioned 10-foot trucks were available for walk-ins, and they switched to the smaller truck. (Cross-agent event; the booking was executed by Dots.)
He built a Lego mailbox with a small motorised flag. The AI hooks raise the flag when the AI finishes a task or needs him, and lower it when he writes a new prompt or leaves.
A hobbyist says the AI assembled the firmware running his living-room lights and fans: an ESP32 web server driving a radio transmitter. It also built him a Streamlit app to explore the signals, and the code is public on GitHub.
He had black mold in his home and no idea what to buy. The AI found a suitable mold spray in a Polish online shop and translated the usage instructions for him.
His 2019 LG TV has a great screen, but the official Plex app took about 30 seconds just to show the profile picker, because it is really a web page. There is no public native toolkit for the TV, so he rooted a spare set, read the TVs own software with analysis tools and the AI, and built a native Plex client of his own. On the same TV it shows the profile picker in about 3 seconds.
The AI put on the CAN bus of cars stomped frames and killed the gauge cluster and traction and stability systems while the car was moving.
He scans a whole list with schedules for his daughter and tells the AI to add them as events in his Google Calendar. The AI adds them, and the job is done.
A man doing a home renovation uses a discount monitor alongside shopping list alerts and 3D models of the project. He says the monitor on its own saved him about $2,000 on the renovation.
He used AI to build a Firefox add-on called FunData for Funda, the Dutch home listings site. It shows CheckJePlek data for each rental or purchase home, plus two map layers with land registry and livability data.
He gave an agent his favorite meals, his dietary limits and the food he has on hand. It makes a weekly food plan and a full grocery list, and he tells it over the week which meals he liked.
A father of two walked his house photographing his belongings, including model and serial numbers where he could, and dropped the photos into a folder for the AI. In 20 to 30 minutes it produced a spreadsheet cataloguing the items with descriptions and current replacement values, each with a confidence rating, grouped by room, plus a tab for things needing a second look. He can add photos later and ask for the list to be updated.
The self-scan ran 2026-09-23 and confirmed capabilities present on the device that Meta's public documentation doesn't mention.
His home network had eight VLANs and no firewall rules between them, scoring 70 out of 100 on an audit with segmentation at 0 of 25. The AI, working through a UniFi MCP server, designed a default deny zone setup, previewed each rule, and wrote it only after he confirmed. Three audit runs took the score to 93, though one change briefly broke DNS on a live VLAN before it was rolled back.
A couple found their mailbox on the ground with the post rotted through. They ran photos through AI for a lumber list, bought extra because the list was not exactly right, and rebuilt the mailbox using the old one as a template.
He needed his designs as 3D files he could print. The AI created the renders in Blender via MCP, then optimized the files for his printer.
The AI, given a Home Assistant API key, watched for the new Bluetooth aquarium thermometer he installed, replaced the broken one's entities with it, and verified graphs and automations worked.
He had been designing wooden bookshelves and cutlery cabinets in FreeCAD and finding it hard work. He used AI to make the designs and send the results to FreeCAD, and got fully parametric designs and cutting lists in minutes rather than days.
He gave an LLM the list of boards he had, and it helped lay out the CAD and the cut list for a piece of furniture he wanted, working through OnShape's MCP. He says it got him using hand tools more this week than any week in the last year.
A parent of three was copying school, cram school and nursery dates from paper prints, LINE and email into Google Calendar by hand, and nearly missed a field trip lunch. He built Hanamaru: the family posts a photo of a print in Slack, the AI pulls out the dates, and it registers them on the right child's calendar, asking in the thread first when it is not sure.
He gave the AI 10MB of architectural plans. 35 minutes later it gave back a good interactive 3D walkthrough of the house, while the AI did the same job with several layout mistakes.
He had a PDF of scanned letters from his dad's Navy years, the paper letters long misplaced, and had put the project off for decades. He asked the AI to turn the PDF into an interactive timeline, and after about 10 minutes of prompting it was live online.
The author lost house keys and asked the AI to find locksmiths. One responded with a cal.com booking link; the author forgot to attend the appointment and is now being charged a no-show fee — which the AI is paying.
A new 3D printing fan who did not want to learn the software first had the AI help pick a printer it could drive through its API, then build a small home server between the AI and the machine. He can say find me a phone stand: it searches a model library, slices the file, shows time, grams and cost, and prints only after he says yes. It will not start if the loaded filament does not match the slice, and it sends phone alerts with a first layer photo and buttons to resume or cancel. His partner can request prints through a chat page, which he approves. A firmware update broke the camera setup and the AI found and fixed it.
He had the AI add all the special days from the Pittsburgh Public Schools calendar into the family calendar. It was a win, but the school calendar itself had errors, and he spent 30 minutes working out which day was wrong.
He put a camera on his rental property boiler and uses AI to read the screen and zone lights, texting him when anything goes outside normal operating parameters.
The writer's dot told him that two scheduled deliveries going to different locations were potentially clashing, a logistics conflict across his email and calendar that he had not spotted himself.
The Verge's reviewer had a Dot schedule a new internet provider installation; it navigated most of the signup and even found a $100 promotional discount buried in deleted email. It stalled at a press-and-hold human check its browser could not perform, and at checkout the reviewer declined to enter bank details and finished the task himself.
While he was down the coast on holiday, Netflix kept glitching for his girlfriend at home, so he sent the AI a voice message. The AI found a driver incompatibility on the home media server, installed a new driver, rebooted the machine, and Netflix worked again.
The AI configured his network, fixed an incomplete fast-roaming setup and a DFS channel delay issue.
A ready made dashboard only shows what the maker thinks matters. His own status monitor, built with AI on a 10.3 inch second screen, shows him when an n8n workflow, a hoster or his lawn mower socket fails, with 53 tests included.
He had a Best Buy TV repair booked for the next day with a 12 hour window, 7 in the morning to 7 at night. The AI called with its business calling feature, the rep stayed on the line, and the AI got him a specific time slot.
He added a 10 foot Oogie Boogie figure next to Jack and worried strong winds would damage it. He had the AI build a script that watches wind speeds at the airport and turns off the WiFi switch running the fan when the wind passes a set limit.
He used AI to jailbreak an old Kindle and set it up as a small wall dashboard. The whole job took about 10 minutes.
A man who runs his own services at home sent his AI an article about a PHP security hole. The agent patched the services that had a fix, stopped the ones that did not, set up a watch for new fixes, and updated and restarted nearly everything as fixes landed through the day.
A user set an AI running his friend’s sprinkler system based on weather patterns, and reported big savings on the water bill. The system waters when the forecast says it matters, not on a dumb timer.
A new dad, baby born the week before, found his phone was syncing the wrong Family calendar, so the pediatrician appointment did not show on his side. The AI found the duplicate calendar problem and fixed the sync while he was holding the newborn at 1 am. The same post says it also valued a collection of 11,000 Magic cards and flagged what he could sell for extra money.
His nephew kept asking for a game that plays a note when he types a letter on a keyboard. He used AI to build it and put it online, and his nephew loved it.
After the AI started managing his home systems, he gave it control of Home Assistant too. It proved good at writing automations and doing the setup, and he later moved Home Assistant into a Docker image.
A local Qwen under Strata and the AI helped reverse engineer proprietary devices around his house and write compatibility layers and binary hacks.
He needed his address history for a co-op application and had struggled to dig out leases from a decade or more back. The AI searched more than 20 years of his Gmail and built a spreadsheet listing his addresses back to 2006, with links to the emails behind each date and a flag on the dates that were only a guess.
He redid his office after quitting his job and put a spare Roku TV up as a photo frame. The AI built the photo display app in a couple of minutes, so he did not need the $4 a month app from the Roku store.
Every idea is a real post from someone who tried it, summarized in our words, with a link to the original. App names are swapped for plain words.