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.
Since the end of August he has added three new CI servers to his own setup. The AI wrote each NixOS setup and he checked the work, so now he can rebuild any of the servers from scratch in minutes.
The AI takes his scanned documents and books them straight into Money, his accounting software.
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.
Writing Rust code faster than state of the art by asking agents to make the code faster.
A Reddit user asked the AI to catch up on unfiled vet bills. The AI searched the email backlog, found five bills totalling $1,504.72, and submitted them to the pet insurer.
He asked his agent Roby to find the WhatsApp number of his car insurer. Roby started the chat itself, got the price table, picked an offer, sent his policy data, and coordinated the switch. He only opened WhatsApp himself to pay the payment link.
He needed quotes for a seasonal tire swap, with the winter set already on rims, and did not want a day of phone tag. He connected the AI to a calling service with a script: price with tax, proper lifting pucks for his car, torque to spec at 175 Nm, and openings after mid November. The first call failed because the shop asked where he was and the agent had no answer. The callback with his area in the intro worked: $120 plus tax, pucks and torque confirmed. Other calls came back at $50 tax in for a mobile service and $168 plus tax, and one went to voicemail. Every call produced a full transcript. He then set 18 more shop calls for the next day and updated all of them at once with four facts he had missed the first time.
Spent half a day switching GPT-5.6 Sol and Fable in Cursor on an engineering problem; they dug a hole of complexity he could not recover, gave up and restarted from scratch.
On a Tuesday, Fulton's the AI warned him that Sunday would open near freezing, connecting a forecast swing to his calendar without being asked. The practical point was jackets for the children before their Sunday activity.
The AI set up a complex Home Assistant routine with lights and motion sensors from a description, working when he walks past the sensor.
Saving his Bluesky threads as Markdown files for Obsidian was a chore he kept doing by hand. He asked the AI to build a converter, and the finished tool is up on his site for anyone to use.
Konik gave the AI a list of low-urgency wants she didn't want to overpay for. Over time it built a table of what normal prices in her area look like and flagged good deals, netting her a $25 armchair and ottoman that fit her reading nook.
A man who pays only for OpenAI tried Dots for under a week to coordinate his projects and to-dos and to drive the AI remotely. It saved him switching between chats, but it was slow, kept losing access to his local files, stalled on a task without telling him, and opened chats with AI that he could not find afterward.
The AI reverse-engineered his laser cutter's G-code dialect and camera controls in 1 to 2 hours; now it places and cuts jobs with power and speed settings and diagnoses problems from scans.
Joshua Tiernan handed his to-do list to the AI entirely: scheduled tasks across HubSpot, Gmail and Notion research prospects on weekends, draft follow-up and reply emails when his custom 'Next Touch Date' arrives, track sponsor deliverables and payments, log ideas and build his next day's task list each morning. Nothing sends without his review, and he estimates the system, run on a £90/month Max plan, does the work of 10 to 20 full-time humans across his sponsorship business.
For the better part of a year an AI agent detected AI-related stories for his hcker.news feed and raised about 20 a day for his judgement, before he replaced the workflow with a fine-tuned classifier.
He runs about 15 apps and sites and could not keep track of what needed doing. He connected the AI to his app through MCP, and the AI now creates tasks and subtasks, writes notes, and shows him what is needed by priority and due date.
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.
The AI reverse engineered the Linux version of DaVinci Resolve to add AAC support; the patch is published and works.
Over the weekend he had the AI set up a daemon on his Mac mini that syncs all of his Mastodon posts to his Obsidian vault. Now he can search his own posts on all of his devices.
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.
The AI’s Comet browser shopped Amazon inside users' own signed-in Prime accounts — loading pages, comparing products, filling carts and completing purchases at the user's direction — until Amazon sued, arguing the access was the user's permission but not Amazon's authorization. A district court enjoined Comet in March 2026; on August 4, 2026 the Ninth Circuit vacated the injunction, holding the agent is a tool and the user is the one buying, while leaving terms-of-service claims live.
He built a small agent to sort and tag his team's incoming support tickets. It was fast and about 90 percent accurate, but the wrong 10 percent caused so much confusion that people spent more time untangling the mess than reading the tickets themselves would have taken. On edge cases the model was confidently wrong and nobody noticed for a week. They scrapped it after a month and went back to tagging by hand.
He is a photographer who found typing alt text for every photo slow going. He built a small agent on his own n8n server: he sends a photo through an Apple shortcut, the agent reads the picture with a local Qwen model, adds the camera data, and saves the finished alt text in an Obsidian note, ten photos at a time.
He runs AI against real client bookkeeping, sorting thousands of PDFs into folders for about 40 companies. Before any file moves, the system writes a log line with the current path, the destination, a sha256 hash of the contents and the path to put it back. Nothing is deleted; files go to a dated quarantine folder, and a file can only be cleared when its hash already exists elsewhere.
Ported a Linux game binary from Cortex-A7 to Allwinner A133P in about 2 hours using AI, with no source.
He ran AI on the task with 6 agents at a time, about 75 agents in total over about 6 days. They produced transcriptions of tens of millions of handwritten images and indexed them all.
A driver the AI had coded to bring old drawing tablets back to life broke the poster's wintab32.dll file, even though the app it was based on never touches that file. Fixing it took 5 hours.
The MBI Deep Dives author writes that the AI tracks a set of data points he cares about and reports daily, including GPU prices, the App Store ratings race between the AI, the AI and the AI, and insider buy/sell transactions in stocks he owns. He describes using roughly 200 million tokens a week on a $20/month plan.
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.