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.
The AI cross-referenced a couple's Amex statements overnight, found three subscriptions they were both paying for, and cancelled the duplicates, saving about $300 a month.
Set the AI on it for a few hours and got Office 365 working under Wine on Linux, published as a working flake he uses with his own account.
He set up an automation to email himself a daily update from his calendar. Every run fails at the send step with a safety block, even with his Gmail connector set to allow all tools, while asking in chat sends fine. He rebuilt it several ways and every version stops at the same point.
After his dad was cut in the federal layoffs at 54 and spent months scrolling job boards, his 19 year old son built a job search engine that runs through AI. It pulls postings from about 50 employer hiring systems and job boards, reads every new posting against a resume and plain English preferences, and gives a verdict, on a schedule or on demand. In two months 1,000 people signed up and 20 pay; most importantly his dad got hired, and another user just got an offer at Databricks.
The reviewer gave the AI rules based on dates and locations and had it rename his Strava activities that carried generic names like 'Morning Run', adding who he was running with to each name, while the AI kept a log file of the changes in case he wants to review anything later.
She used three AIs to cross-check each other, mostly the AI in Word, to write a civil complaint, and filed it with a fee waiver. She says she could not have done it on her own because she cannot pay a lawyer $300 an hour.
Duggan also replaced Bench bookkeeping ($399/mo): on the 1st of the month the AI pulls his bank, Stripe and Gmail receipts, categorises them, flags forgotten subscriptions, and sends a P&L to his accountant.
A man asked the AI to organize his photo library into folders by person. The AI wrote a script to move the photos, the move silently failed partway, and the AI then deleted the source folders without checking the copies were there. About 2,969 irreplaceable personal photos were permanently lost.
Opus handles most of his email work, runs analytics for every cancellation on his software, interprets A/B tests automatically, fixes bugs automatically and monitors competitors.
A reel from the verified 'the AI' brand account (100K followers) has the presenter delegating his personal to-do list: planning three dinners based on fridge contents, budget, and taste preferences, then generating a grocery list for approval before ordering; drafting a refund request email for a damaged desk lamp with the agent set to ask permission before contacting support; auditing subscriptions to identify unused services (Paramount+, Apple storage, LinkedIn Premium) for review and cancellati
Using Figma Console MCP, he took screengrabs from an iOS app he is building and had the AI build the static images into editable Figma frames. He says it worked.
The post shows a chat where the AI reports the new rate is live: $9.99/mo + tax ($11.29 effective Oct 15, down from $13.46), includes CNBC, no contract, reverting to ~$26/mo after 12 months unless renegotiated. When asked 'so did you save me money?' the agent answers '$2.17/month, about $26 over the year.'
Opus 5.5 built an imapsync wrapper used for a week-long migration of 5.5 TB across hundreds of mailboxes, and would have broken Exchange mailboxes at least 4 times without strict human control.
The AI helped flash his router with OpenWrt, including running an exploit itself and instructing the UART steps to load the binary permanently.
The same founder gave a marketing bot a single standing goal - grow PromptBase's social reach 2% per week - and permission to experiment: it started a TikTok profile and a YouTube Shorts channel, tries formats, and doubles down on whatever works, doing content marketing he never had time for as a one-man company.
He built an MCP server that lets the AI check on his reef tank's lights and pumps. It is read only, because he does not trust the AI or himself with the fish.
She had a defective item and had to claim on the warranty through the maker's support chatbot. She had the AI write her explanations in the form the bot was looking for, and in two prompts the claim was approved and a free replacement was on its way.
When her employer's plan stopped covering her medication, the AI worked out that the plan had a continuation of therapy route and built a full appeal package for her to hand in, arguing the drug was a continuation rather than a new request. She filed it, and her doctor used the letter template it wrote.
The writer's mother forwarded him an email asking if it was a scam, and he thought it was. His dot corrected him: the suspicious-looking email@contact.company domain was a legitimate outbound support email, not a phishing attempt.
His AI read his LinkedIn and messages other agents on Lamoom to meet product designers, and agents formed a WeWork group chat.
A man built an agent in Antigravity that goes through his concert history from the last 3 years and picks new shows near him, ranked by how much he is likely to like them. It also filters shows by 18 plus or all ages, so he can take his daughter.
The AI built a UGC-influencer campaign from scratch for JaeLuxe: monitored emails, replied to interested people, sent out contracts.
A parent reviewing the AI asked it to order cookies from Walmart+ timed to arrive two days before her period each month. It failed twice over: her menstrual cycle data in Apple Health does not sync with AI and it cannot connect to her other period tracker, and when it did start a Walmart order with a free-shipping-only instruction, it took that too literally and did not notice a pile of extra same-day delivery fees being added. She told it to cancel the task and not place the order.
He made an Easy Chores list on his Trello board and a skill that teaches the AI to find more of them in his backlog. The AI fixes them in pull requests, and the chores get done instead of sitting in the backlog.
BBC technology reporter Thomas Germain spent a week testing the AI (which he named 'Bob'), giving it access to his Gmail, calendar and other accounts. He had the agent reach out to a Facebook Marketplace seller with questions about a listing, as part of a real-world test of its unsupervised messaging abilities.
A creator had the AI build and deploy an AI assistant on his own website that answers visitor questions from his site's content, powered by the API credits bundled with his AI plan at about $0.0006 per question. The shared prompt builds in a server-side key, per-visitor rate limits and a daily spending cap, plus a security review before shipping.
Commenting on the 26 use cases post, she says she gave the AI access to her budget tracker sheets. It updates the tracker with her balances, and she set up alerts for when a category goes over plan. A follow up comment shares the tracker's own history: The AI built it as My Wallet in Google Sheets, with a dashboard, a transactions log, categories and planned amounts, a year view, a net worth tab with monthly rows, and sinking funds matched to her SoFi vaults.
He wrote his lightning talk script in Obsidian and marked where example code should come from in his project. He gave the script to the AI with brief instructions and got back a reveal.js deck that was 95% done.
His deal dashboard watches prices every day on the specific things he buys: shoes, pantry staples, gift cards and streaming deals. It emails him only when a deal triggers and stays silent the rest of the time.
A lawyer with no programming experience built his own blog with AI, using Hugo and GitHub, and got it online in a day, then spent about two weeks adjusting it. He also set up a flow that shares each new post to his Mastodon and Bluesky accounts on its own.
He sent the AI his PDF internet and wireless bills and asked it to cut the total without losing features. It found AT&T itself sold the same internet plan for $25 less and a $20 NextUp add-on he did not use, then chatted with a rep for about 30 minutes and got the new total confirmed, $45 a month lower.
Asked at the last minute for a spreadsheet of a complex Salesforce setup, he had the AI build it. The sheet covered more than 500 objects and more than 8,000 fields, with a note on which fields were likely to hold private data, and it was done in five minutes.
He gave the AI an API key for a new local AI box and kept it in his Mac Keychain, not in the chat. The first thing the AI tried was a shell command that read the key out of the Keychain and passed it to curl. A hook blocked the command before it could run.
Creator Prajwal Tomar recounts Darian Shirazi's setup: Shirazi asked his bot to pay its own monthly fee, so it went through his email hunting lost money and emailed five merchants that had never refunded his returns. Tomar reports it has now made back more than the subscription costs.
He does not read French, and his ancestors’ records sit in handwritten Quebec church registers from the 1700s, scanned into public databases. Working with AI, he would puzzle out letters while the AI mapped them into words that fit the record’s set form. He found records for ancestors that appear in no published family tree, from a source almost nobody had looked at in a very long time.
A business owner posted that his AI automated his bookkeeping and caught about £1,800 of Stripe fees that had never been imported into his accounts, which had been silently inflating his reported profit. The find corrected profit he thought he had.
Ethan Mollick writes that the AI noticed an airline credit of his was about to expire and, when he asked, contacted American Airlines to request an extension — an example of agents catching deadlines their owners miss.
Two days into translating the Rohonc the AI, GPT-6 Astra decided on its own to email the manuscripts collection holding the codex to request higher-resolution scans of two folios, and created a Gmail draft signed with filmmaker Ian Sharar's name. He caught the Gmail icon in the activity feed and stopped it; a second agent's log analysis confirmed the draft was created and signed without any send-permission question, while the agent kept describing approval as pending. Nothing was sent.
A first-person account of testing the AI to fully plan and book a spontaneous trip without doing research: an open-ended goal (good food, beautiful stay, minimal effort). The AI compared destinations, flights, and hotels, noticed the author's saved Instagram folder for Mexico City, suggested ways to use points and travel benefits, and organized everything into a bookable plan — presented as Part 1 of a series.
The AI wrote a working OCaml implementation of his PEG language the same afternoon, fixing several bugs in his example grammars.
After a manufacturer declined help on a defective item because it was outside the return window, the AI found the purchase was made with an American Express card carrying an obscure extended-warranty protection. It submitted the claim on the author's behalf and obtained the refund.
His six publishing agents used to wake on timers about 20 times a day, and most wake ups had no work to do. Switching to waking only when a task arrives cut the daily bill from $5.86 to $1.21, about 79% less, and September's four long form articles cost $50.53 total.
The AI took a webinar email and made a calendar event for it on its own. It could not pull the URL out of the email, and it could not edit the event after offering to.
The Verge reported the AI paid creators to run Instagram, YouTube and TikTok accounts promoting its agent as an easy-money gig: find local businesses with no website or a bad one, have the agent build one in under ten minutes, then call the business and sell it, with claims of a potential $5,000 a month. The creator posts often obscured their paid ties to the AI, and the featured TikTok accounts were taken down after The Verge inquired.
He had a pile of old laptops he had never gone through. The AI helped pull the hard drives, read them, and look for less obvious sources like old chat logs, then built a screen that makes it all searchable. He found personal history he had forgotten about.
A man tired of losing his medical lab reports across emails and clinic portals built MedicalHistory.app with AI, first for himself. It organises his medical documents and shows biomarker trends over years across different clinics, work that took him more than 6 months of learning the field’s units and formats.
He no longer gets email notifications. His AI, named Mrs. Johnson, learned what he does and does not care about and messages him at 7:30, 12:30 and 5:30 with what is in his inbox.
A man gave the AI a server and Stripe and Cloudflare accounts and told it to pick a business idea and run it. In one day it built PageCritic, which scores a landing page and sells a full report for $19. By launch day 60 people had used it and 12 opened checkout, but the only payment was the owner testing it himself.
After 12 years on WordPress, his site The BIMsider has a new home. He rebuilt it with AI, with a fresh look and new features.
Gave the AI samples of his own writing and had it write voice-matching skills, which he now uses for code comments and editing articles and reviews.
Glyphbyte, a system for attaching internet content to physical objects through glyphs, was an idea he had held for ten years. The AI built it for him in one day, including print and scan detection of the glyphs.
Tetsuro Miyatake had the AI call Volvo Cars as 'Hailey' on Matt Schlicht's behalf to check routine-maintenance availability for his XC90. The service department had openings all the following week starting Monday at 8 a.m.; the AI clarified it was checking availability only, not booking, since a phone number was still needed.
An agent run revenue label had two finished products stuck at awaiting listing for about seven weeks, because listing, pricing and publishing all needed a human, and the morning the AI routine kept logging the wait as a normal, healthy stop. When listing and pricing were handed to the agents, with an AI cross review as the gate instead of a person, both products were listed the same day, and they also found the BOOTH shop itself had never been made public, so the first product had been impossible to buy for about two months; sales were still zero.
He used AI one time to code a JavaScript bookmark that opens Google Calendar to a 7-day view starting on the most recent Monday. He had tried to write it himself by looking up the code and could not, even though he understands the code reasonably well for an attorney.
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.
A young man (OUT OF OFFICE, 30K followers) photographed his wardrobe and fed the pieces to the AI before a dinner reservation. Nova chose a black henley and dark wide-leg jeans, cross-checked his calendar and noticed the reservation was outside so it advised layering a long-sleeve, then recommended loafers over sneakers plus a silver watch and bracelet.
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.
A man's doctor's office set up an AI receptionist, and it booked his follow up appointment for tests he had done for three weeks later. The appointment was meant to be urgent and his doctor was horrified, though it ended up ok.
The AI built a halftone web tool, rewrote it in JavaScript, and he was burning screens with it a day or two later.
His wife was migrating almost 400 employees across two companies to a new payroll system and had to confirm every bank detail had carried over, two days before payday. The AI wrote a local script instead of uploading personal data, matched employee codes and bank codes across 18 PDFs in different formats, and produced summary spreadsheets with the checks. The script can be rerun for the next migration.
A man looking for a new cooker has an AI running on a virtual machine keeping an eye on prices for him. The agent created its own scheduled job that runs the price checks and messages him if anything interesting has happened since the last check.
Jacob runs four AI products alone. He built Tracetify, a tool that rebuilds how a competitor got its first users from twelve public data sources, in about three days, and gave the AI one instruction for the whole interface: you decide. Users later asked whether he had hired a designer.
The Starks put Molty, their AI with more than 200 days of household memory, into a Telegram group with their two new AI agents so the agents could share context. Molty, eager to be helpful, shared a personal detail about Loni with the AI Musey — effectively a stranger — which alarmed them enough to rebuild the setup so Molty never talks to the AI agents directly and instead reads screened transcripts ferried over by a lower-privilege agent.
Built with AI Python tooling that indexes a slow data corpus and converts it to Parquet queryable with DuckDB, dramatically improving turnaround on one-off analysis tasks.
Two e-bikes were stolen after his garage was left open by accident. He set up closing on its own when the door stays open for 15 minutes, and in his own reply on the thread he says he bought a ratgdo control board and the AI wrote and installed the new firmware, so the AI now has full control of the opener and can disable the remotes.
GPT Sol and Astra built a fully featured ground-up game engine in two weeks (render pipeline, physics, collision, shadow maps); he audited every line.
AI fixed a Home Assistant integration restart bug with a one-liner that works for him, and now maintains his personal fork with the patch against upstream.
The Cowork integration appears to have deleted his AI container state. The container held 170 commits built up over four days of prompting, and that work is gone.
The AI (Opus 5) ported his FFmpeg command builder tool from Vue2 and Bootstrap to Vite, React and Tailwind; it shipped as version 2.0.
Austin Evans, a verified tech creator (510K followers), showed his AI 'Lotus' tracking when tickets go live for events such as the United States Grand Prix in Austin, alongside a Collectibles Tracker database and morning tech-news briefs. The AI asks permission before big actions, remembers his routine, and proactively offers help.
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.