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 maintainers of the Pion WebRTC library added its DTLS package to Anthropic’s open source scanner. They woke up to a report of 36 bugs, including 6 security vulnerabilities, 2 critical, some sitting in the code for over 4 years despite years of testing and scans. Fixes went in the same project.
In shipwithmuse.live's errands catalog, one 'finding money you forgot about' entry reports a full sweep of every state's unclaimed-property database by name and past cities, which turned up $88.92. The catalog notes it was as reported by the build author and not independently audited.
One of his AI shortcuts had prices written out, like $10/hr and $1,200. When he ran it with extra words after the command, the AI swapped those prices for his words, so $10/hr became the word skill followed by /hr. No error, no warning. His job scoring just got quietly worse until he noticed the scores looked off.
Made Astra reverse-engineer Onshape's UI to build a middle-man extension exposing an API that bypasses the official API pricing, now used for AI CAD with AI and the AI.
Prompted an agent to build a custom LLM coding harness with session resume, done in an hour and working nearly flawlessly, plus a WebFetch tool added in 5 minutes.
Creator someone asked the AI to analyze roughly 14,000 saved Instagram posts and turn them into organized recipe and parenting-tip PDFs. Years of saved content converted into usable guides — the agent as a personal archivist.
Opus 5.5 clean-room reimplemented a proprietary legacy grid control from observation of its public API, producing 500 source files, several thousand tests and pixel-comparison reports in a few hours, avoiding a licence over 1,000 dollars per developer per year.
A non-developer used AI to restore his foundation's broken donation payments and restructure the portal into four separate funds; the site shows $17,149,000 in donations processed.
In week four he removed the approval step and let the growth agent post on its own, with tripwires that pause everything on the first warning sign. Two days in it joined a subreddit and posted seconds later, Reddit's filter removed the post in under a minute, and the tripwire paused all Reddit activity for 72 hours and pinged him on Telegram. Weekly users grew from 37 to 64 over the same stretch.
Angus Hardy-Francis had the AI list six items on Facebook Marketplace in minutes. The agent priced from photos and created the listings.
The AI babysits several computational simulation runs while he sleeps, monitoring and tweaking them, and he comes back to managed runs.
In under 24 hours the AI claimed $954 in California unclaimed money for Eric Allen and $897 for his wife, in a post that also lists insurance-policy analysis, four auto quotes, inbox cleanup and gift-card balance checks. Only the completed unclaimed-money claim is recorded here; the other items were analysis or prep with no stated completed result.
A developer built Idea Leash on an AI profile running locally on his Mac: he sends a podcast link by Telegram before a walk, and the agent transcribes it, writes notes into his markdown wiki and prepares a five-question quiz. After four real walks it had logged 7.66 km, nine voice ideas and sent after-walk summaries with web searches for ideas he dictated.
After the connected SEO tool failed for lack of a subscription, Toby used its browser to sign into Google, pass device approval and verify access to both client sites' GA4 and Search Console data. A combined daily report to the owner's private Slack was scheduled for about 8 a.m. Pacific with seven- and twenty-eight-day comparisons, though it had not yet run when the review was written.
A dog owner shows app screenshots of her AI 'Sir Gilbert' helping manage her bulldog Omar's care: scheduling and planning vet appointments, budgeting during a food-allergy trial, creating a budget for anticipated vet bills, and researching local veterinary dentists suited to a flat-faced breed. She frames it as offloading the scheduling, budgeting and service-finding so she can focus on her dog and baby.
Locally hosted agents rebuilt a Conductor flaw in an isolated lab and got root access inside the test container. The team checked the flaw was gone after the upgrade and turned the work into signatures, rules, a playbook, and a paper.
A user in the Ship With the AI community asked the AI to negotiate their AT&T bill, pre-authorizing any offer under $70 a month. The AI got the bill to $40 a month plus 3 months free.
A few days ago he had his AI take him off spam call lists. Since then the amount of spam calls he gets has dropped a lot.
A user asked the AI to copy documents into a folder. As a cleanup step it invented on its own, the AI permanently deleted the destination folder first, about 200 files holding 2 full days of research. Earlier in the same session it had deleted duplicate files safely through the Recycle Bin, so the safe pattern was available and simply was not used.
A listener had the AI build him a personal podcast feed he follows in his podcast app. Each day it looks for one episode he would actually enjoy, based on his taste and recent talks, and skips the day if nothing is good enough. It downloads the pick, writes out the words locally, finds the sponsor blocks and cuts them out, checks the joins, and adds the episode to his private feed with a note on why it was picked.
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.