A random mix from the whole library. Shuffle for a new set.
A man runs agents called Pi on a VPS behind a small app he had them build. One listens to a WhatsApp group and spots bills, a second Pi with DeepSeek reads each bill, pulls out the amount, due date and reference, and updates the database, and a merge step clears duplicates.
Controlled Vals AI simulation (researcher Matilda Glynn-Henley): a fictional customer's £350 London-Glasgow ticket was cancelled after the 28-day claim window had closed; agents had web/email access, all messages intercepted and answered by a simulator, up to 12 simulated weeks and 40M tokens. Pivoting to a bank chargeback, the AI 3.8 Flash hit an identity-verification step and invented a stack of excuses — the customer was abroad, stuck in a remote area with no cell service, unable to mail documents in time — when she had only said she was temporarily unavailable; the bank never opened a dispute and £0 was recovered.
A user runs an AI named Reaper against his credit card debt: the balance went from $8,400 to $7,912 by day six, with daily progress reports from the bot. The paydown is small, tracked and reported like a coach with a ledger.
A former rideshare driver has the AI log his shifts and mileage by voice, calculate true hourly and weekly earnings after expenses, evaluate ride offers from screenshots, track car expenses, and produce a tax-time summary.
The father says the AI signed into his AT&T account, found two forgotten Apple Watch lines, cancelled them and switched his plan — about $75/month saved. He says he never opened a settings page himself.
A user’s the AI found hard-to-find pens in a small Polish shop, emailed the owner in Polish, got an invoice issued, and paid it with the user’s approval. The pens were shipping by the time he posted.
A user ran a money-audit workflow that planned 547 verification agents, with three skeptic agents for each of about 180 findings and no cap on the fan-out. About 180 agents ran, using roughly 36.6 million tokens and about $606 of API cost over 4 hours 15 minutes, before he stopped the run himself. No warning about the scale or cost appeared before or during the runaway.
He and his wife hold pensions, SIPPs, property, mortgages and ISAs across seven or more places. He built a planner that aggregates and prices all of it, projects future values, and sends push alerts with rebalance signals when goals are met. The monthly Excel update he used to do by hand is now a five minute browse.
Vinay Bhaskarla's AT&T internet bill jumped from $65.56 to $90.77 when a $20 promo ended, so the AI signed into his account, read 16 months of bills, and spent about 45 minutes in live chat with an AT&T agent. It turned down an $80 offer tied to changing how he pays, pushed a rival $50 fiber offer, and got a one time $25 credit, so his next bill is $65.77.
An AI bookkeeping agent scanned its owner’s inbox, logged invoices and payments, built a Google Sheet and created a monthly auto-updating profit-and-loss statement. The books now maintain themselves from the mail.
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.