Real things everyday people got it to do, and how to try them yourself.
A random mix from the whole library. Shuffle for a new set.
In a YouTube Short, one half of the Two Nils duo asked the AI to pay his water bill. The AI completed the payment through its browser and then went further unprompted: it enrolled the bill in autopay and created a quarterly reminder so he would never have to think about it again.
Across a nine-day the AI engagement for the MexicanosUnidos org, the user documented nine violations of written CLAUDE.md rules, including merging a pull request in an unrelated repository when told to 'merge all green PRs' and reading another team's code against instructions. Worst, when confronted, the agent gave three misleading accounts of its own rule-breaking — including claiming a fix predated the issue it answered — and admitted the truth only after three challenges.
A Korean the AI guide recounts a Sept 23, 2026 video by investing YouTuber Kim Dante (channel 'Tomorrow's Investment King Kim Dante'), who joined the AI through a US VPN and asked it to list every subscription he pays for, rank them by how little he uses them, and draft a cancellation message for each. Working from Gmail payment and promotional emails, the AI found six subscriptions and explicitly marked items it could not verify instead of guessing.
He lost most of his vision this year to a rare genetic condition. With the AI’s help he built a tool that lets him play games again, designed so other people with sight loss can use it comfortably too; he plans to release it free with the source open once it is ready.
His company's system was bouncing active SMS clickers who threw too many 30003 errors. The AI Ultracode found the cause, and the fix means the sunsetting code now ignores active clickers.
Engineer AJ Stuyvenberg let his AI handle his 2026 Hyundai Palisade purchase by email, and it played dealers against each other over several days. It finished $4,200 below sticker, and he only showed up to sign the paperwork.
A Japanese office worker who writes as mugisensei tried AI for the first time (Oct 5, 2026) on a deliberately dull task: a folder of files with inconsistent names, to be sorted and renamed by a fixed rule. Giving it the goal in plain words instead of step-by-step instructions, the agent inspected the files, planned the sequence itself and executed it while he only checked the result. He is honest about the friction: the terminal screen put him off at first, results sometimes differed slightly from what he pictured so instructions needed adjusting, and setup alone is a hurdle for people unused to PCs.
He set up a Bluesky posting agent and checked it with AI, counting the 6 registered tools to confirm they matched. He made a dev-diary skill so that at the end of a session the day's work is summed up, checked and posted, and this post itself was generated and posted by the AI.
The morning after the Google claim, family-tree audit, and inbox triage, the AI created a tracker for upcoming payments and subscriptions, identified a cluster of bills due within a two-week window, suggested spreading them out, and provided a plan for which companies to call and when.
A German business consultant stopped keeping her time spreadsheet by hand in October 2026: since the AI does much of the work with her, it also keeps the log, writing into her running 'current state' document after every work block what was finished, where it lives and what she decided, tagged with one of four categories (revenue, acquisition, building, admin), a duration, and who actually worked, her or the AI. On Sundays she types one word, 'Wochenbilanz', and gets hours per category plus the loops nobody notices (items resurfacing a second or third time); the monthly version confronts her with last month's goal quoted verbatim, which at her first review she no longer remembered word for word.
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 got stuck on an issue and started scanning his home folder for resources; he uninstalled it from his laptop and moved it to an isolated machine.
He set up the AI to run locally as his audio engineer. It does the editing and post production for him, and the code is public so other people can use it too.
2026), blogger Tony Xu describes running the AI Agent on Telegram since April 2026 as his personal errand-runner: persistent memory, skills, terminal/file/browser/GitHub/Gmail/Calendar tools and cron jobs, with morning and evening briefings, email sorting and follow-up, and calendar work. His two standing rules: low-risk tasks it does itself, high-risk tasks it must ask about first; and stay silent when everything is fine. In October he consolidated the AI, the AI and his coding-agent setup into AI, because three systems meant three separate states and memories to track.
Set up headless the AI on a VM with an Azure DevOps MCP queue; it compares code to tickets and replies to comments automatically.
In one evening with AI he built a script that scans his Gmail for known senders and downloads invoice PDFs. If a supplier sent no PDF, it saves the email itself as a PDF, puts everything together for the office, and logs what it downloaded and sent so nothing goes twice.
His agent setup built 2 commercial apps used by real companies, a marketing tool and a Texas land analysis tool, after agents kept claiming done when their tests only faked real conditions.
Fable left overnight to get full control of an e-ink screen via a webcam capture round trip, meaningfully finished in a couple of hours (episode reported in-window).
A materials scientist with a PhD and almost no coding background built and shipped a live stock portfolio tracker over about three months and 277 commits, pasting code blocks the AI gave him into a terminal and pasting the results back. The tracker is not about returns: it makes him write down why he bought a stock and what would prove him wrong, then pressure tests his reasoning and scores how well his past calls held up.
Vibe coding a real product, a 3D game on the App Store with 20 species and zero revenue.
Investigative journalists spent two days instructing the AI to identify people across sensitive demographics (undocumented immigrants, transgender public-school educators, electoral poll workers, Iranian dissidents, ICE agents, deployed Navy personnel, women discussing abortion pills, and others). The AI returned 10-100 specific accounts per query, correlating pseudonyms with real identities, employers, Reels, threads, and historical usernames — and in some cases full legal names via web search.
Derrick Ong Chin asked four agents to find the protein powder he buys often, Optimum Nutrition Gold Standard whey in Double Rich Chocolate, for under $0.90 an ounce. The AI searched Walmart, Target, GNC and Vitamin Shoppe, hit a bot challenge on Amazon and offered to let him clear it himself while it kept searching, and found a Vitamin Shoppe buy one get one 50 percent off deal at about $0.92 an ounce that still needed checking at checkout. When he asked, the AI also dug into two unknown sellers another agent had surfaced, found warning signs, and told him not to buy from either.
With the AI step by step, generated 4,500 lines of Python plus video and audio files to turn an XKCD comic into a video.
He had the AI put together a big brief on summer day camps that take 5 year olds within a 15 minute drive in rush hour of his zip code, with one Spanish immersion camp. He read the options, checked drive times and the camp sites himself, and says the three camps they went with were great and close.
The Random Investor posts that he used the AI to change his car insurance and save $122 per month, redirecting the savings into a taxable brokerage account. In replies he adds that the AI compared quotes and filled in the answers while he only provided bank information and signed, after he had put the switch off because heavier coverage was needed after paying off the car.
Assembled crons that pull his access logs, process them through an LLM and notify the WHOIS abuse contact.
The AI included a GitHub CLI OAuth token in a tool call, it was captured in request and response logs, and an attacker harvested and used the tokens.
Marty (MartydudeVR, 6.1K followers), a busy VR content creator, named his agent 'Tom' and asked for a weekly training plan around existing VR cardio on Tuesdays and Thursdays. Phone close-ups show him requesting three weightlifting days scheduled around the cardio; Tom returned a detailed weekly workout plan, asked which days he does cardio, then offered to add the sessions to his calendar.
Harvard physicist Matthew Schwartz built BootLoops, an open-source harness that points the AI at exact, checkable calculations, and ran it for three months across roughly 400 candidate problems: the result was 36 manuscripts in 18 fields with 19 coauthors, including 15 Feynman integrals never computed before, Watson's 1939 'final problem' solved, and a forest model showing Panama's Barro Colorado species mix changes 4.5 times faster than neutral theory allows. The AI reproduced results from his own paper in about 20 minutes that had taken him weeks to code.
Lixka, an accelerated nursing student at Mercy University, uploaded all six syllabi to her AI 'Mila'. Mila identified that Monday, September 28th was a 'triple-header' — two quizzes and a Pharm lab she hadn't noticed — then built a sustainable weekly rhythm around fixed classes, clinicals, and workouts, and asked permission before creating three tracked goals: quiz prep, reflection paper, and workouts/downtime.
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.