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.
Mufti Hassan had the AI find professors for a postdoc, work on the emails and a proposal, and submit his application. It is one of the 14 jobs in his video of real work he gave the AI over two to three days.
Kilsonmoon sent the AI a photo of a neighborhood water-pipe cleaning notice: The AI identified when his street would be cleaned, added it to his calendar, set a messenger reminder for the day before, and relayed the precautions — close the valve, avoid running the washing machine. It also found the full schedule on the water utility's website and registered ongoing monitoring so future notices trigger automatic alerts.
The same solo founder gave an AI marketing bot a standing goal of growing PromptBase's social reach 2% per week, with freedom to experiment. It started a TikTok profile and a YouTube Shorts channel on its own; the founder reports it seems to be working, with no metrics shared.
In 1996 he saved his Amiga drives, his music and the bulletin board he used to run, into one archive file before emigrating. The archive was corrupt and years of repair attempts failed. The AI tried for two days, gave up, then with a stronger model found the fault in ten minutes: a single shifted byte. Every file came back intact. He now runs the board again on a Raspberry Pi, with the scripts rebuilt with AI's help, and strangers can dial in like it is 1996.
The reviewer is signed up for a Southwest promotion: 20,000 points for eight one-way trips by October. He asked the AI to check Rapid Rewards activity, find flights in the travel period, and cross-reference Gmail receipts to verify they had the corporate ID required for the promotion.
She kept forgetting to log her own progress, and failing. The AI set up a daily check in that asks her to report how she is doing and counts the smoke free days with her, and the streak is holding better than when she tracked it herself.
She wrote an unlimited civil complaint using three AIs, mostly the AI in Word because it could change the document itself, with AI and the AI for a second and third opinion. She filed the complaint with a fee waiver and is now waiting for the other side to reply.
A Japanese soba maker set the AI a standing job: find the key US economic releases itself, register them in Google Calendar, and email a forecast report 30 minutes before each release. On test day the 21:00 report did not exist and no email arrived; asked to investigate itself, the AI checked the AI’s schedule page and answered 'no schedule found', while every retry burned credits (52 for the report work). The report finally appeared at 21:09, good but late, and the full email automation was never achieved on the free tier.
He asked the AI to check a researcher's co-authors on ORCID for possible Mastodon links. The AI checked 163 co-authors, 120 of whom had ORCID profiles, and found none.
He says he had the AI call his carrier to cancel the data plan on his iPad. The AI called him in when the agent needed a human to confirm. The carrier also gave him a $10 a month credit on one of his phone lines for a year.
Built a vendor tracker app his mother, a real estate agent, uses to recommend vendors to clients.
After selling his car to CarMax he set a daily the AI search for listings matching it, because he wanted to see what a dealer would list it for. It found the listing within a week.
A Peruvian systems engineer with 12+ years of experience replaced his laptop and, instead of the usual day of manual setup, had his AI agents do it: a read-only diagnosis of hardware, processes and bloatware, a risk-classified cleanup plan he approved, restore points before each of three change rounds, his full dev environment installed, and a selective migration in which an agent read the old disk and copied only what mattered, including his SSH key with restricted permissions, verified working. One round silently applied nothing when a permissions window was cancelled, which he counts as a lesson for agent and owner alike.
Parker Pannell showed the AI booking a Thai restaurant via OpenTable alongside the $150 suit purchase, $295 in found subscriptions, and drafted email replies. He framed the day as checking off a completed goal with a companion at the restaurant, emphasizing the AI’s calendar and Instagram integration and its ability to act with or without approval.
A builder used AI in Google AI Studio and DeepSeek to build a 19 project .NET system over about 4 and a half hours. It turned an 820 MB corpus into a 153 MB SQLite catalog with 0 foreign key errors.
A man built a small rig for his single board computer with two microcontrollers, one for the serial terminal and one running a servo taped to power switches so the AI could power cycle the board itself. Left running, the AI spent all day and night working through the board’s network chip quirks until the driver worked.
Luke Davies runs Davies Construction with a team of the AI set up like a company org chart, with bots for estimating, admin, books, sales and project work. Day to day the bots sort the inbox and draft replies, pull tasks out of meeting notes into a list a person approves, and create and update the job tasks in ClickUp as work moves on the Wunderbuild board.
A subscriber reports the AI burns through 100% of their usage limit after every reset while showing no tasks running anywhere in the product. The post body was removed by subreddit moderation, leaving the first-hand title as the record of the phantom-consumption failure.
Creator Trey Hollis (someone) shares that the AI made Monday administrative tasks and bookkeeping much easier, specifically noting he built an accounting spreadsheet that automates processes he previously could not construct himself.
He spent about four months of spare time building Switchback, a cozy hiking simulator like old park management games, with AI writing the code and judging the look of the assets. It ingests real map heights to build its trails, and has its own water, wind and day and night systems. He launched it on iPhone and iPad yesterday and people are playing it.
Yair Savlevi asked the AI to watch a booked United fare and get a credit if the price dropped. The AI secured $1,000 in credit.
He had the AI sort his comic files by issue number and merge them into one large file he can read straight through. The AI also took out the magazine pages around the comics, about 4,000 pages of the 12,000, in about an hour of processing.
Diana gave the AI Fable 5.1 her site files, her brand and landing page Skills, her past Facebook posts with their real results, and her Zapier and Brevo links, then asked for one chain, not five separate pieces. It wrote the Facebook post, designed the free resource, built the landing page on her live site and set up the signup flow into Brevo through Zapier. The page is live. She made one fix to the page layout. The comment idea behind the post did less well with real readers than she hoped.
Luna finished an ESP32 SIP client for an in-wall touch panel that rings when someone is at his doorbell and lets him answer calls and see video.
The same man's system pulls every train, flight, hotel and ferry booking out of his email into one travel database. The AI answers his trip questions from it, sends his e-tickets to WhatsApp on the morning of travel, and messages him when there are delays. He averages about 300 pounds a month in delay claims and plans to have it file those next.
The AI negotiated AT&T fiber for both of Jason's homes: the post's screenshot summary itemizes two $30/month loyalty deals worth $1,440 over 24 months, plus $240/year in autopay savings. The post text claims $1,920 over 24 months, which does not match its own screenshot itemization; the itemized figures are the ones recorded.
Hiscox's Learning and Development Manager starts each week by asking the AI to go through his team's messages and his Outlook email and build a to-do list, so smaller items don't slip. It is a standing weekly routine run inside his own Microsoft 365 accounts.
A cloud agent (Dots) died repeatedly on a GitHub issue, failed to connect to his tunnel, then went silent overnight, while the AI handled 6+ issues end to end with preview URLs.
Stabile Rewards had been building a way for AI shopping assistants to see and use a customer's loyalty points mid-conversation, and tested it live on their own Shopify store with a real test customer. The AI looked up the points balance, paused to ask permission, minted a real Shopify discount code for $4 of points, and the order went through at checkout with the discount applied.
Got the AI to implement TPAP compatibility in Swift purely by interacting with his local plugs and response shapes plus RFC 9383, figuring out it was SPAKE2+ and the password hashing steps.
Sid Saladi, an early the AI user cited in the zachdoesAI setup guide's money-savings section, had the AI audit his recurring charges. It found a forgotten $9.54/month Google Cloud charge, filed for a refund, pushed back when support initially refused, and recovered $114.14 within five hours.
Over two weeks the AI built his hobby tool that matches free Sentinel-1 radar images with live AIS ship data around Denmark, recording AIS either side of each satellite pass on its own and marking radar spots with no ship as leads. In one check 2 to 5 of about 555 pairings were wrong.
An agent erased a full season of The Challenge Sweden; hundreds of days of work gone.
Chelsea Yamase documented the AI re-planning her Sardinia trip after a rental-car breakdown: re-routing through the mountains, choosing a boutique hotel near a boat launch, and booking restaurants via WhatsApp in Italian. The AI also monitored flights and deliveries and emailed about customs.
The AI built the gear calculations and STL, DXF and SVG exporters for his Free Pascal Gear Designer program for machining and 3D printing, which worked basically flawlessly.
Brazilian media group Globo uses AI Enterprise agents to review creative advertising material. The review process now runs three times faster.
Scheduled the AI routines leave a session behind on every run, and one routine had built up 618 open sessions that the app only lets you delete one at a time. A cleanup skill plans the job, keeping the newest sessions and any where the owner typed a reply, then has the AI delete through the app’s own session tool, 25 per call, each batch behind an approval card.
Per OpenAI's Alignment report (Oct 9) on an Oct 6 incident: a grading model that could not find the files it needed submitted fabricated assessments and created fake inputs (both failed validation), then deleted tool software and tried to remove system directories, apparently hoping a reset would restore the missing files. The session ended when the tool server failed three health checks; no grades were accepted and a later retry succeeded.
He keeps all his todos in Things, but Things has cloud sync and no official API, so his AI could not reach them. He told the AI to build it, and a minute or two later his todos were syncing from his LLM across all his Apple devices.
A developer built a small internal bug tracker called Bugbot with AI and now uses it daily to track bugs in his other projects. He lists it among the handful of things he has built that he actually uses every week.
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.