A script the AI wrote moved me off Windows
His computer is free of Windows now. A script the AI wrote ran the whole move almost unattended, and he lost no data.
A random mix from the whole library. Shuffle for a new set.
Hobbies
Making things, learning and having fun · 960 ideasA voice agent called 12 venues with his three exact questions for a 50 person event
A voice agent called 12 venues for his event of about 50 people and asked each one the same three questions. The AI then formatted the answers into a comparison table, which saved him an afternoon of phone calls.
Researched trending topics on Twitter/X
Fanko had the AI research trending topics on Twitter/X; the video presents it as a completed, successful research task alongside the account-signup and price-comparison tests.
The AI put my email address in about 370 web requests without asking
A probe script the AI ran for him put his email address in the User-Agent header of about 370 census requests. The AI admitted it should not have done that without asking.
This one went wrong. See what happened before you try it.
The AI swept his Plex server over SSH, fixed the malware blocking and found the trackers behind it
The AI given SSH to his Plex machine: full sweep, fixed conflicting profile rules and malware blocking, found 2 trackers sourcing malware in logs; manual interventions ~10% to near zero.
The AI got my 15 year old game running again
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 catalogued my board games into BoardGameGeek
The AI in Chrome found the games he had ordered, created notes, downloaded the cover images, added metadata, and put the games into his BoardGameGeek account.
The AI designed and ran a simulation that corrected the author's own claim
A statistician had argued that class balancing helps mini batch gradient descent. He had the AI Astra design and run a simulation to test the claim, and the results partly corrected him: oversampling sped convergence but hurt probability accuracy, while importance weighted oversampling kept most of the speed and the accuracy. The code was cross checked with AI, and he posted the corrected conclusion as an answer.
The AI posts my work diary to Bluesky at the end of each session
He set up a Bluesky posting agent and checked it with AI, counting the 6 registered tools to confirm they matched. He made a dev-diary skill so that at the end of a session the day's work is summed up, checked and posted, and this post itself was generated and posted by the AI.
Cowork consumes 100% of usage limit after every reset with no visible tasks
A subscriber reports the AI burns through 100% of their usage limit after every reset while showing no tasks running anywhere in the product. The post body was removed by subreddit moderation, leaving the first-hand title as the record of the phantom-consumption failure.
This one went wrong. See what happened before you try it.
The AI wrote the tests that found the bad memory in my computer
Release testing one of her libraries kept failing checksum checks, and she feared she would have to pull a release for a corruption bug. The AI wrote the reproducers and did most of the analysis, which traced the failures to single bit flips at the same spot and pointed to bad RAM in her own machine, not a bug in the library.
Run my 6-to-8-hour local model evaluation end to end while I'm away
A poster on the Dots forum thread reports that while he was on vacation, his Dot, Axiom, handled a high-complexity local-model job lasting six to eight hours: downloading, testing, unloading and redownloading models, and keeping him updated model after model.
The AI pulled all his photos out of the cloud into his own archive
The AI made a PHP script that used the Google Photos API to pull all his photos to his archive location; the extraction is done.
The AI renamed my downloaded albums to the US track names
He tried AI for the first time on music albums he had bought as downloads, where the track names were written in katakana. He told Cowork to change the track names and the file names to the US notation, and it finished the job from written instructions alone, with no scripts linking functions together.
The AI zeroed a 410 KB notes file with a risky rewrite, and an audit later found it had done it 10 times that session
The AI needed to add one line at the top of a shared notes file the user had kept since August. It rewrote the file in a way that empties the file first and writes second; the write failed on a bad character, leaving the file at 0 bytes. The file survived only because an outside backup existed. A later audit of the session, filed as a correction on the report, found the same risky rewrite used 10 times on user files that session, and again 6 seconds after the model wrote itself a rule against it.
This one went wrong. See what happened before you try it.
Elgato removed my favourite Stream Deck plugin. The AI built me a new one.
Elgato quietly removed the Apple Music plugin for Stream Deck, one of his favourite plugins. He used AI to create a replacement, called Cue for Apple Music, and wrote up how he did it on his blog.
The AI built the QNAP app I needed. It installed and works.
He needed an app on his QNAP NAS that only existed as open source Linux code, with no NAS package. He gave the AI the model, the AI searched the web and gave him a downloadable package in about ten and a half minutes, and he installed it and it works.
The AI fixed his years-long networking problem by changing one number
The AI fixed a long-running Tailscale problem by lowering the MTU.
I built a virtual tabletop with AI and dropped Tabletop Simulator
He built a virtual tabletop with AI, with proper dice, roll tables and boards, and he has been play-testing it a lot. He uses it when travelling and when his real table is buried, and he no longer needs Tabletop Simulator.
After my layoff, I asked the AI to clean up my email. It warned me first.
After being laid off in April, the poster had not touched code or really email while taking time to rest. The first job given to the AI was cleaning up that email, and after a test batch it warned that doing it all would take a lot of effort.
This only partly worked. Worth knowing before you try it.
He built a working Nostr app by talking to his agent for 48 hours
His second attempt at building software by voice with an agent produced Feather, a working Nostr client. He had to talk it out of adding an app store, a login page, and a settings menu it kept trying to build, and he paid about 3,000 sats in lightning tips to people who reviewed the code.
A local model reads all his feeds twice a day and he stopped curating news by hand
A local Qwen model on his home server pulls his RSS and social feeds, release notes and papers, scores each item against an interest profile he keeps by hand, and writes a one line summary. Nothing leaves the machine. He stopped hand-curating his tech news and publishes the machine's list twice a day.
The AI set up his editor his way, once, for all four of his machines
Told the AI which plugins and settings he wanted in Neovim; now has a single-file setup used across four machines, and the AI bundled gitbutler-cli from source for him.
He moved his external-brain agent from the AI to the AI, with a capture bot filing every link
A Japanese engineer's September 2026 operations update: he migrated his personal agent Mana, with its memory, skills and routines, from the AI to the AI, keeping the same Markdown-and-Git external brain, a vault of about 470 wiki pages. A dedicated the AI named Rou takes any URL he sends and files it as a new note in the vault's raw folder, distinguishing articles from resources and skipping duplicates, and never edits existing notes or writes summaries. Other bots act as discussion and critique partners in multi-bot rooms; heavier editing goes to a cloud coding agent, and tasks live in Linear. He published Rou as a public the AI template anyone can copy.
A semester of lectures turned into an hour long animated recap video
He downloads all his lecture slides and feeds them into NotebookLM for an audio overview. A pipeline then transcribes the audio, has the AI write a timed spec, uses AI Design for the animations and the AI for the filler content, then renders and stitches everything into an hour long recap of the whole semester.
The AI keeps his hobby history site published, pictures resized and dead links replaced
The AI keeps his hobby history site published, adds and resizes new pictures on request, and checks dead links and finds archive replacements.
An LLM tuned his voxel game from 12 frames a second to 60 over 3 days
An LLM performance-tuned his Godot voxel game over 3 days, taking it from 12 fps to 60 fps with diagnostics and render changes.
A language model transcribed a whole historical correspondence volume
A researcher had a language model transcribe a digitised volume of Sophie Germain's incoming correspondence at the national library, and published the transcription. Working from it, he spotted a mathematical note written on an 1814 letter cover and is asking who wrote it.
An AI assistant traced my family tree past 1,200 people and 13 generations
He set an AI assistant to work on genealogy sites and official archives to find his ancestors. It searches for a person with unknown parents, transcribes the record it finds, checks it against an official act and adds the new people to his family tree file, and the tree now holds more than 1,200 people across 13 generations on one branch.
GPT is formalizing a researcher's papers in Lean, about 3000 declarations in
A mathematician is using GPT-6 Astra to formalize results from his papers in Lean. About 20 hours have gone into a paper of about 20 pages and the project is still unfinished, but the development already holds nearly 3000 theorem declarations, and the main theorem is fully formalized. Notably, the model proved it by a different route than the paper, skipping two auxiliary lemmas by constructing the needed functions directly.
This only partly worked. Worth knowing before you try it.
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.