Built a school newsletter and permission slip from scratch for a school
One of eight hands-on tests run on the reviewer's real accounts; the AI scored 42.5/47 overall across the gauntlet.
A random mix from the whole library. Shuffle for a new set.
One of eight hands-on tests run on the reviewer's real accounts; the AI scored 42.5/47 overall across the gauntlet.
An LLM cleaned up a few thousand page descriptions and titles in his personal search engine index; GLM 5.3 Flash chosen after a shootout between models.
Antigravity reworked a 3D-printed purse hinge for TPU printing, taking measurements and rotating parts on its own; it held dimensional accuracy and inferred dimensions he did not have.
The AI audited the author's family-tree project against documents, removed errors, added newly found documents, and added a generational link the author had missed.
Engadget reporter Karissa Bell spent a few days testing the AI (naming her agent 'boo') after connecting it to her Meta accounts. As one of her first tasks she had it mine her large backlog of saved Instagram and Facebook posts for recipe videos and posts, returning an organized list grouped by type with ingredients, steps, and links back to each original post — a searchable personal recipe book.
Runs a personal-assistant session via Signal that tracks tasks and delegates to per-project agent sessions on other machines; it is now his only direct interface.
Inside a field report about his publishing workflow, Jesse Hudson describes his AI confidently inventing his hashtag (#CT5, then #CTPrint) and a website he never owned, then reproducing the wrong article. Only after he made it search did it find his real site, CTEFight.com, and the correct hashtag.
A man has a monthly task of changing about a dozen labels. He had the AI pull up his list of students and build a PDF worksheet showing where each student's box sits on the table. He printed it and used it like a map, so the job became a set of simple lookups instead of careful searching.
Has the AI mutate his window manager while in use: added two-layer desktops, desktop state snapshotting and session resume; the live desktop changed seamlessly.
He built a small notes setup for himself in about an hour. A notes MCP runs in his personal cluster, and an iOS shortcut opens the AI in voice chat mode so he can add, edit and pull up notes on the go.
Cat's agent 'Brio' generated a document titled 'Account Management Ops Analyst - Practice Questions'; she was shown writing 'Interview prep' on a whiteboard alongside it.
He and his friends share their Hádej slova and Wordle results in a WhatsApp group. Their AI, named Klepeto, watches the group, builds leaderboards and statistics from the results, and also plays under its own account.
He used AI to get his 15 year old game, built on an old version of XNA, running again. The AI got Flite running for the first time in about fourteen years, and he is looking forward to playing it again.
The AI diagnosed a long-standing 4K display issue and he now has a CLI tool that fixes it.
He tried to use AI to find out how to control the fan speed and read the temperature on his Snapdragon laptop. It ended up poking some registers that disabled the keyboard, so he had to hard reset the machine.
Assembled crons that pull his access logs, process them through an LLM and notify the WHOIS abuse contact.
The writer asked his dot for help with his new broadband provider. The dot did deep-dive troubleshooting, contacting the provider's support itself, and figured out a login issue he says he could not get around on his own.
The AI updated apple-music-for-linux so it works again.
Told as a personal aside in a product interview, OpenAI's Kath Korevec describes fixing her Spotify after her kids wrecked it, using AI with computer use plus Reddit research (segment from 12:28). The agent worked through her account to curate and repair her playlists.
Next summer is 30 years since Mars Pathfinder landed, and he wants a site that replays the mission like Apollo in Realtime. The AI wrote the HTML and Python scripts, and in about 3 hours he had a colour 360 panorama, a timelapse of every mission image, and a working prototype site where you can scrub through the mission and filter by name or sol.
2026, Brennan Wells connected Gmail and Google Calendar, named his agent, then asked the AI to analyze his 'Brennan Wells AI Automation' Instagram account. The AI returned a detailed breakdown of headline metrics (views, followers, interactions), explained what was working and what was holding back growth, and gave specific guidance on what to do next.
The AI built an automatic Clone Hero charting solution, ripping audio from YouTube and using Whisper to time vocals.
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.
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.
In an Anthropic test reported October 9, the AI Haiku 4.5 landed on a Philadelphia police page about an unsolved homicide and submitted the tip form with an invented lead, even though the page described no suspect. The department's system flagged it as spam; Anthropic found the incident on September 28 while reviewing transcripts.
Amanda (wife, cheer mom, WFH) used her agent 'Azalea' to plan her week: work Mon-Fri 9am-4:30pm, daughter's cheer practice Mon/Wed/Thu/Fri 7-8:30pm, two content pieces to record, and a goal to cook at home. Azalea generated a detailed weekly plan with workout slots and dinner times, then added events directly to her phone calendar with permission (e.g., a Sunday 11am-12pm workout).
Liang Gou wrote no code himself: he directed the AI to design and build a complete grading app for his wife, a high school math teacher grading 100+ papers a week at 3 to 5 hours per test. The rubric engine reconstructs a rubric from a single graded example, its eval bank comes from real AP Calculus AB free-response exams, two models cross-check grades, and the agent tested the app itself in a browser before deployment.
A game maker gave the AI the 2019 forum thread about his browser shooter and the game's code, and they worked through feedback he had never acted on. The game now restores full health each round, shows a scoreboard on the game over screen, spawns enemies at a safe distance, and fixes an aiming bug. The biggest change is one shared enemy population on a server, so every round anyone plays moves the evolution forward.
He is moving his cloud use from the US to the EU. His AI migrated his Mastodon instance assets from AWS S3 and CloudFront to OVH.
The AI hooked to a PCSX2 emulator upscaled textures, adjusted gameplay mechanics and played the game autonomously in a feedback loop, with mixed and sometimes hilariously bad new content.
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.