A random mix from the whole library. Shuffle for a new set.
John Merphy made $600 selling two short client research reports: The AI compiled the research, compared sources and formatted the docs; he reviewed and delivered. Part of $1,150 total that month.
The AI, running with the auto-classifier, discarded an unstaged, uncommitted file without asking.
A man fed an expanded spec sheet into AI Sites and got a working booking website for his rehearsal space in 30 minutes. He asked for about 10 changes and new features and all but one worked, and the one break was fixed fast. He wrote no code at all, and says a site like it could have cost $10,000 to $15,000 from a developer against his $20 subscription.
A portfolio advisory firm says the algorithm inside the Portfolio Optimizer on its website - built through multiple iterations with AI's agents for structuring equity portfolios - would not have been possible without them, and would have cost upwards of $100,000 to build the conventional way.
The AI and the AI built his Obsidian plugin, listed in the Obsidian Directory with 65 users, alongside shell scripts, Alfred workflows and Elisp functions for his own setup.
The AI ported CADO-NFS to GPUs and orchestrated a fleet on scavenged idle capacity, running about 30 GPU-years over 10 days to factor RSA-260.
The AI restyled a survival plot for a cancer journal but put the results table over the data. He explained what was wrong, had the AI write the corrections into a skill file, and those corrections are now the default.
Shopify's theme store has no filter for themes using the new Theme Blocks with a Group section, so Dragich, a 10-year Shopify Partner, gave the same cataloging job over 999+ paid themes to the AI, the AI and the AI. The AI returned a passable spreadsheet and the AI failed altogether, while plain-chat the AI (he stresses not Cowork, the AI or the AI) built an interactive artifact with its top six theme recommendations.
Opus 5 pointed at his own app created a similar exploit to a known file-upload attack in 3 minutes, showing the app was vulnerable.
He turned his 15,000 saved Wispr Flow dictations into a test set for the local Whisper model behind his AI Agent. The agent made find and fix rules from the words Whisper kept mishearing, tested each rule against the full history, and got hard words right far more often: 99 of 240 up to 183 of 240 on old clips, and 19 of 45 up to 43 of 45 on sentences he read out loud.
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.